archive.Joe Lamantia.com

JoeLamantia.com complemented my formal professional work in technology, design, product, and strategy, beginning with the early(ish) Web moment of the middle 90s.

For approximately 15 years, beginning about 2000, the site shared practice-related tools, methods, frameworks, industry and academic publications, professional presentations, and evolving perspectives (after 2006, genuinely raw thinking out loud mostly happened via Twitter).

The final postings publish industry analysis on the then-emerging field of data science, with a broader frame of the expansionary category of analytically-driven business products and services.  My roles at the time emphasized product strategy for B2B software applications and B2C platforms powered by predictive models built from collections of business, consumer, and research data – composite assets newly recognized as Big Data.  I described this as  the machine intelligence space, to clarify the focus on new technology and product development outcomes, and distinguish the broader category of AI.

From 2016 onwards, with ‘software eating the world‘, my professional roles shifted to leading scaled / scaling product development groups, with charters emphasizing innovation powered by the expanding stable of human-centered technology disciplines: information architecture, interaction design, user experience, content strategy, design research, product strategy (still not well-articulated…).  With broad success and growth within the business context (read, steady buyers), the questions shifted from foundational — e.g. codifying ‘What is User Experience?’, and shaping ‘How does it even happen?’ — to operational  — ‘How is this done better at scale?  In new channels?  For the entire business? With customers around the world?’

To answer these basic ‘get it done’ questions on crafting human-informed products daily within large business contexts, the cross-border communities for product development, technology, and media spun up a healthy circuit of professional gatherings,  and a layer of complimentary social forums.  Conversations that originally took place via small group gatherings and niche news groups or listservs, shifted to Big Conferences, and Big Social Platforms.

In that landscape, there was less to share directly in the blog format.  Also, there was the rest of life: family, home, community.

Then in 2018, after a series of minor maintenance and administration incidents, that show how the social Web and the entire Internet environment was changing to a regime of financialized surveillance capitalism, and algorithmically amplified predation, there was no ‘there’, there.  JoeLamantia.com went dark, as far as sharing my work was concerned.  The domain was doing a different job, for different audiences. and stayed that way.

Before I decided to focus fully on looking ahead and making new things for the new spaces of the early Web, I’d planned to study history, media, and technology – essentially looking in the other direction, as a scholar.  I *almost* did a PhD at U. Chicago or Pitt (thanks to both programs for seeing potential and offering opportunity).  This path not taken taught me the deep value of a historical perspective, especially when you’re considering where to go next, and how to get there.

Now, almost exactly ten years since the last original post in June of 2014, following a modest technical reanimation effort, I’m happy to offer a restored archive version of JoeLamantia.com.  It’s not *everything* that was written, said, or shared — but it’s most of what mattered.  We’re back.

To move forward, we’ll be reflecting on some of the “practice-related tools, methods, frameworks, industry and academic publications, professional presentations, and evolving perspectives” shared, to assess and learn from them by looking in both directions.

Thanks for your consideration: then, and now.

Concept models are a powerful tool for articulating the essential elements and relationships that define new or complex things we need to understand.  We’ve previously defined empirical discovery as a new method, looking at antecedents, and also comparing and contrasting the distinctive characteristics of Empirical Discovery with other knowledge creation and insight seeking methods.  I’m now sharing our concept model of Empirical Discovery, which identifies the most important actors, activities, and outcomes of empirical discovery efforts, to complement the written definition by illustrating   how the method works in practice.

In this model, we illustrate the activities of the three kinds of people most central to discovery efforts: Insight Consumers, Data Scientists, and Data Engineers.  We have robust definitions of all the major actors involved in discovery (used to drive product development), and may share some of these various personas, profiles, and snapshots subsequently.  For reading this model, understand Insight Consumers as the people who rely on insights from discovery efforts to effect and manage the operations of the business.  Data Scientists are the sensemakers who achieve insights, and create data products, and analytical models through discovery efforts.  Data Engineers enable discovery efforts by building the enterprise data analysis infrastructure necessary for discovery, and often implement the outcomes of empirical discovery by building new tools based on the insights and models Data Scientists create.

A key assumption of this model is that discovery is by definition an iterative and serendipitous method, relying on frequent back-steps and unpredictable repetition of activities as a necessary aspect of how discovery efforts unfold.  This model also assumes the data, methods, and tools shift during discovery efforts, in keeping with the evolution of motivating questions, and the achievement of interim outcomes.  Similarly, discovery efforts do not always involve all of these elements.

To keep the essential structure and relationships between elements clear and in the foreground, we have not shown all of the possible iterative loops or repeated steps.  Some closely related concepts are grouped together, to allow reading the model on two levels of detail.

For a simplified view, follow the links between named actors and groups of concepts shown with colored backgrounds and labels.  In this reading, an Insight Consumer articulates questions to a Data Scientist, who combines domain knowledge with the Empirical Discovery Method (yellow) to direct the application of Analytical Tools (blue) and Models (salmon) to Data Sets (green) drawn from Data Sources (magenta).  The Data Scientist shares Insights resulting from discovery efforts with the Insight Consumer, while Data Engineers may implement the models or data products created by the Data Scientist by turning them into tools and infrastructure for the rest of the business.  For a more detailed view of the specific concepts and activities common to Empirical discovery efforts, follow the links between the individual concepts within these named groups.  (Note: there are two kinds of connections; solid arrows indicating definite relationships, and for the Data Sets and Models groups, dashed arrows indicating possible paths of evolution.  More on this to follow)

Another way to interpret the two levels of detail in this model is as descriptions of formal vs. informal implementations of the empirical discovery method.  People and organizations who take a more formal approach to empirical discovery may require explicitly defined artifacts and activities that address each major concept, such as predictions and experimental results.  In less formal approaches, Data Scientists may implicitly address each of the major concepts and activities, such as framing hypotheses, or tracking the states of data sets they are working with, without any formal artifact or decision gateway.  This situational flexibility is follow-on of the applied nature of the empirical discovery method, which does not require scientific standards of proof and reproducibility to generate valued outcomes.

The story begins in the upper right corner, when an Insight Consumer articulates a belief or question to a Data Scientist, who then translates this motivating statement into a planned discovery effort that addresses the business goal. The Data Scientist applies the Empirical Discovery Method (concepts in yellow); possibly generating a hypothesis and accompanying predictions which will be tested by experiments, choosing data from the range of available data sources (grouped in magenta), and selecting initial analytical methods consistent with the domain, the data sets (green), and the analytical or reference models (salmon) they will work with.  Given the particulars of the data and the analytical methods, the Data Scientist employs specific analytical tools (blue) such as algorithms and statistical or other measures, based on factors such as expected accuracy, and speed or ease of use.  As the effort progresses through iterations, or insights emerge, experiments may be added or revised, based on the conclusions the Data Scientist draws from the results and their impact on starting predictions or hypotheses.

For example, an Insight Consumer who works in a product management capacity for an on-line social network with a business goal of increasing users’ level of engagement with the service wishes to identify opportunities to recommend users establish new connections with other similar and possibly known users based on unrecognized affinities in their posted profiles.  The data scientist translates this business goal into a series of experiments investigating predictions about which aspects of user profiles more effectively predict the likelihood of creating new connections in response to system-generated recommendations for similarity.  The Data Scientist frames experiments that rely on data from the accumulated logs of user activities within the network that have been anonymized to comply with privacy policies, selecting specific working sets of data to analyze based on awareness of the shoe and nature of the attributes that appear directly in users’ profiles both across the entire network, and among pools of similar but unconnected users. The Data Scientist plans to begin with analytical methods useful for predictive modeling of the effectiveness of recommender systems in network contexts, such as measurements of the affinity of users’ interests based on semantic analysis of social objects shared by users within this network and also publicly in other online media, and also structural or topological measures of relative position and distance from the field of network science.  The Data Scientist chooses a set of standard social network analysis algorithms and measures, combined with custom models for interpreting user activity and interest unique to this network.  The Data Scientist has predefined scripts and open source libraries available for ready application to data (MLlib, Gephi, Weka, Pandas, etc.) in the form of Analytical tools, which she will combine in sequences according to the desired analytical flow for each experiment.

The nature of analytical engagement with data sets varies during the course of discovery efforts, with different types of data sets playing different roles at specific stages of the discovery workflow.  Our concept map simplifies the lifecycle of data for purposes of description, identifying five distinct and recognizable ways data are used by the Data Scientist, with five corresponding types of data sets.  In some cases, formal criteria on data quality, completeness, accuracy, and content govern which stage of the data lifecycle any  given data set is at.  In most discovery efforts, however, Data Scientists themselves make a series of judgements about when and how the data in hand is suitable for use.  The dashed arrows linking the five types of data sets capture the approximate and conditional nature of these different stages of evolution.  In practice, discovery efforts begin with exploration of data that may or may not be relevant for focused analysis, but which requires some direct engagement to and attention to rule in or out of consideration. Focused analytical investigation of the relevant data follows, made possible by the iterative addition, refinement and transformation (wrangling – more on this in later posts) of the exploratory data in hand.  At this stage, the Data Scientist applies analytical tools identified by their chosen analytical method.  The model building stage seeks to create explicit, formal, and reusable models that articulate the patterns and structures found during investigation.  When validation of newly created analytical models is necessary, the Data Scientist uses appropriate data – typically data that was not part of explicit model creation.  Finally, training data is sometimes necessary to put models into production – either using them for further steps in analytical workflows (which can be very complex), or in business operations outside the analytical context.

Because so much discovery activity requires transformation of the data before or during analysis, there is great interest in the Data Science and business analytics industries in how Data Scientists and sensemakers work with data at these various stages.  Much of this attention focuses on the need for better tools for transforming data in order to make analysis possible.  This model does not explicitly represent wrangling as an activity, because it is not directly a part of the empirical discovery method; transformation is done only as and when needed to make analysis possible.  However, understanding the nature of wrangling and transformation activities is a very important topic for grasping discovery, so I’ll address in later postings. (We have a good model for this too…)

Empirical discovery efforts aim to create one or more of the three types of outcomes shown in orange: insights, models, and data products.  Insights, as we’ve defined them previously, are discoveries that change people’s perspective or understanding, not simply the results of analytical activity, such as the end values of analytical calculations, the generation of reports, or the retrieval and aggregation of stored information.

One of the most valuable outcomes of discovery efforts is the creation of externalized models that describe behavior, structure or relationships in clear and quantified terms.  The models that result from empirical discovery efforts can take many forms — google ‘predictive model’ for a sense of the tremendous variation in what people active in business analytics consider to be a useful model — but their defining characteristic is that a model always describes aspects of a subject of discovery and analysis that are not directly present in the data itself.  For example, if given the node and edge data identifying all of the connections between people in the social network above, one possible model resulting from analysis of the network structure is a descriptive readout of the topology of the network as scale-free, with some set of subgraphs, a range of node centrality values’, a matrix of possible shortest paths between nodes or subgraphs, etc.  It is possible to make sense of, interpret, or circulate a model independently of the data it describes and is derived from.

Data Scientists also engage with models in distinct and recognizable ways during discovery efforts.  Reference models, determined by the domain of investigation, often guide exploratory analysis of discovery subjects by providing Data Scientists with general  explanations and quantifications for processes and relationships common to the domain.  And the models generated as insight and understanding accumulate during discovery evolve in stages from initial articulation through validation to readiness for production implementation; which means being put into effect directly on the operations of the business.

Data products are best understood as ‘packages’ of data which have utility for other analytical or business purposes, such as a list of users in the social network who will form new connections in response to system-generated suggestions of other similar users.  Data products are not literally finished products that the business offers for external sale or consumption.  And as background, we assume operationalization or ‘implementation’ of the outcomes of empirical discovery efforts to change the functioning of the business is the goal of different business processes, such as product development.  While empirical discovery focuses on achieving understanding, rather than making things, this is not the only thing Data Scientists do for the business.  The classic definition of Data Science as aimed at creating new products based on data which impact the business, is a broad mandate, and many of the position descriptions for data science jobs require participation in product development efforts.

Two or more kinds of outcomes are often bundled together as the results of a genuinely successful discovery effort; for example, an insight that two apparently unconnected business processes are in fact related through mutual feedback loops, and a model explicitly describing and quantifying the nature of the relationships as discovered through analysis.

There’s more to the story, but as one trip through the essential elements of empirical discovery, this is a logical point to pause and ask what might be missing from this model? And how can it be improved?

 

Related posts:

One of the most compelling outcomes of our strategic research efforts over the past several years is a growing vocabulary that articulates our cumulative understanding of the deep structure of the domains of discovery and business analytics.

Modes are one example of the deep structure we’ve found.  After looking at discovery activities across a very wide range of industries, question types, business needs, and problem solving approaches, we’ve identified distinct and recurring kinds of sensemaking activity, independent of context.  We label these activities Modes: Explore, compare, and comprehend are three of the nine recognizable modes.  Modes describe *how* people go about realizing insights.  (Read more about the programmatic research and formal academic grounding and discussion of the modes here: https://www.researchgate.net/publication/235971352_A_Taxonomy_of_Enterprise_Search_and_Discovery) By analogy to languages, modes are the ‘verbs’ of discovery activity.  When applied to the practical questions of product strategy and development, the modes of discovery allow one to identify what kinds of analytical activity a product, platform, or solution needs to support across a spread of usage scenarios, and then make concrete and well-informed decisions about every aspect of the solution, from high-level capabilities, to which specific types of information visualizations better enable these scenarios for the types of data users will analyze.

The modes are a powerful generative tool for product making, but if you’ve spent time with young children, or had a really bad hangover (or both at the same time…), you understand the difficult of communicating using only verbs.

So I’m happy to share that we’ve found traction on another facet of the deep structure of discovery and business analytics.  Continuing the language analogy, we’ve identified some of the ‘nouns’ in the language of discovery: specifically, the consistently recurring aspects of a business that people are looking for insight into.  We call these discovery Subjects, since they identify *what* people focus on during discovery efforts, rather than *how* they go about discovery as with the Modes.

Defining the collection of Subjects people repeatedly focus on allows us to understand and articulate sense making needs and activity in more specific, consistent, and complete fashion.  In combination with the Modes, we can use Subjects to concretely identify and define scenarios that describe people’s analytical needs and goals.  For example, a scenario such as ‘Explore [a Mode] the attrition rates [a Measure, one type of Subject] of our largest customers [Entities, another type of Subject] clearly captures the nature of the activity — exploration of trends vs. deep analysis of underlying factors — and the central focus — attrition rates for customers above a certain set of size criteria — from which follow many of the specifics needed to address this scenario in terms of data, analytical tools, and methods.

We can also use Subjects to translate effectively between the different perspectives that shape discovery efforts, reducing ambiguity and increasing impact on both sides the perspective divide.  For example, from the language of business, which often motivates analytical work by asking questions in business terms, to the perspective of analysis.  The question posed to a Data Scientist or analyst may be something like “Why are sales of our new kinds of potato chips to our largest customers fluctuating unexpectedly this year?” or “Where can innovate, by expanding our product portfolio to meet unmet needs?”.  Analysts translate questions and beliefs like these into one or more empirical discovery efforts that more formally and granularly indicate the plan, methods, tools, and desired outcomes of analysis.  From the perspective of analysis this second question might become, “Which customer needs of type ‘A’, identified and measured in terms of ‘B’, that are not directly or indirectly addressed by any of our current products, offer ‘X’ potential for ‘Y’ positive return on the investment ‘Z’ required to launch a new offering, in time frame ‘W’?  And how do these compare to each other?”.  Translation also happens from the perspective of analysis to the perspective of data; in terms of availability, quality, completeness, format, volume, etc.

By implication, we are proposing that most working organizations — small and large, for profit and non-profit, domestic and international, and in the majority of industries — can be described for analytical purposes using this collection of Subjects.  This is a bold claim, but simplified articulation of complexity is one of the primary goals of sensemaking frameworks such as this one.  (And, yes, this is in fact a framework for making sense of sensemaking as a category of activity – but we’re not considering the recursive aspects of this exercise at the moment.)

Compellingly, we can place the collection of subjects on a single continuum — we call it the Sensemaking Spectrum — that simply and coherently illustrates some of the most important relationships between the different types of Subjects, and also illuminates several of the fundamental dynamics shaping business analytics as a domain.  As a corollary, the Sensemaking Spectrum also suggests innovation opportunities for products and services related to business analytics.

The first illustration below shows Subjects arrayed along the Sensemaking Spectrum; the second illustration presents examples of each kind of Subject.  Subjects appear in colors ranging from blue to reddish-orange, reflecting their place along the Spectrum, which indicates whether a Subject addresses more the viewpoint of systems and data (Data centric and blue), or people (User centric and orange).  This axis is shown explicitly above the Spectrum.  Annotations suggest how Subjects align with the three significant perspectives of Data, Analysis, and Business that shape business analytics activity.  This rendering makes explicit the translation and bridging function of Analysts as a role, and analysis as an activity.

Subjects are best understood as fuzzy categories [http://georgelakoff.files.wordpress.com/2011/01/hedges-a-study-in-meaning-criteria-and-the-logic-of-fuzzy-concepts-journal-of-philosophical-logic-2-lakoff-19731.pdf], rather than tightly defined buckets.  For each Subject, we suggest some of the most common examples: Entities may be physical things such as named products, or locations (a building, or a city); they could be Concepts, such as satisfaction; or they could be Relationships between entities, such as the variety of possible connections that define linkage in social networks.  Likewise, Events may indicate a time and place in the dictionary sense; or they may be Transactions involving named entities; or take the form of Signals, such as ‘some Measure had some value at some time’ – what many enterprises understand as alerts.

The central story of the Spectrum is that though consumers of analytical insights (represented here by the Business perspective) need to work in terms of Subjects that are directly meaningful to their perspective — such as Themes, Plans, and Goals — the working realities of data (condition, structure, availability, completeness, cost) and the changing nature of most discovery efforts make direct engagement with source data in this fashion impossible.  Accordingly, business analytics as a domain is structured around the fundamental assumption that sense making depends on analytical transformation of data.  Analytical activity incrementally synthesizes more complex and larger scope Subjects from data in its starting condition, accumulating insight (and value) by moving through a progression of stages in which increasingly meaningful Subjects are iteratively synthesized from the data, and recombined with other Subjects.  The end goal of  ‘laddering’ successive transformations is to enable sense making from the business perspective, rather than the analytical perspective.

Synthesis through laddering is typically accomplished by specialized Analysts using dedicated tools and methods. Beginning with some motivating question such as seeking opportunities to increase the efficiency (a Theme) of fulfillment processes to reach some level of profitability by the end of the year (Plan), Analysts will iteratively wrangle and transform source data Records, Values and Attributes into recognizable Entities, such as Products, that can be combined with Measures or other data into the Events (shipment of orders) that indicate the workings of the business.

More complex Subjects (to the right of the Spectrum) are composed of or make reference to less complex Subjects: a business Process such as Fulfillment will include Activities such as confirming, packing, and then shipping orders.  These Activities occur within or are conducted by organizational units such as teams of staff or partner firms (Networks), composed of Entities which are structured via Relationships, such as supplier and buyer.  The fulfillment process will involve other types of Entities, such as the products or services the business provides.  The success of the fulfillment process overall may be judged according to a sophisticated operating efficiency Model, which includes tiered Measures of business activity and health for the transactions and activities included.  All of this may be interpreted through an understanding of the operational domain of the businesses supply chain (a Domain).

We’ll discuss the Spectrum in more depth in succeeding posts.

Related posts:

Big Data, Language of Discovery,analytics, business_analytics, discovery, language_of_discovery, sensemaking, sensemaking_spectrum

Last year, I had the pleasure of collaborating on a paper with Tony Russell-Rose and Stephann Makri that builds on and extends our work to understand and articulate a framework for discovery needs and activities – what we refer to as the Language of Discovery – showing examples of concrete application and use.

It’s been a while in coming, but I’m happy to say the complete paper ‘Defining and Applying a Language for Discovery’ – is available now.

I’ve reproduced the complete text of the paper below, and there’s also a pdf for download.

Abstract

In order to design better search experiences, we need to understand the complexities of human information-seeking behaviour. In this paper, we propose a model of information behaviour based on the needs of users across a range of search and discovery scenarios. The model consists of a set of modes that users employ to satisfy their information goals.

We discuss how these modes relate to existing models of human information seeking behaviour, and identify areas where they differ. We then examine how they can be applied in the design of interactive systems, and present examples where individual modes have been implemented in interesting or novel ways. Finally, we consider the ways in which modes combine to form distinct chains or patterns of behaviour, and explore the use of such patterns both as an analytical tool for understanding information behaviour and as a generative tool for designing search and discovery experiences.

1 Introduction

Classic IR (information retrieval) is predicated on the notion of users searching for information in order to satisfy a particular ‘information need’. However, much of what we recognize as search behaviour is often not informational per se. For example, Broder [2] has shown that the need underlying a given web search could in fact be navigational (e.g. to find a particular site) or transactional (e.g. through online shopping, social media, etc.). Similarly, Rose & Levinson [12] have identified the consumption of online resources as a further common category of search behaviour.

In this paper, we examine the behaviour of individuals across a range of search scenarios. These are based on an analysis of user needs derived from a series of customer engagements involving the development of customised search applications.

The model consists of a set of ‘search modes’ that users employ to satisfy their information search and discovery goals. It extends the IR concept of information-seeking to embrace a broader notion of discovery-oriented problem solving, addressing a wider range of information interaction and information use behaviours. The overall structure reflects Marchionini’s framework [8], consisting of three ‘lookup’ modes (locate, verify, monitor), three ‘learn’ modes (compare, comprehend, evaluate) and three ‘investigate’ modes (explore, analyze, synthesize).

The paper is structured as follows. In Section 2 we discuss the modes in detail and their relationship to existing models of information seeking behaviour. Section 3 describes the data acquisition and the analysis process by which the modes were derived. In Section 4 we investigate the degree to which the model scales to accommodate diverse search contexts (e.g. from consumer-oriented websites to enterprise applications) and discuss some of the ways in which user needs vary by domain. In addition, we explore the ways in which modes combine to form distinct chains or patterns, and reflect on the value this offers as a framework for expressing complex patterns of information seeking behaviour.

In Section 5 we examine the practical implications of the model, discussing how it can be applied in the design of interactive applications, at both the level of individual modes and as composite structures. Finally, in Section 6 we reflect on the general utility of such models and frameworks, and explore briefly the qualities that might facilitate their increased adoption by the wider user experience design community.

2 Models of Information Seeking

The framework proposed in this study is influenced by a number of previous models. For example, Bates [1] identifies a set of 29 search ‘tactics’ which she organised into four broad categories, including monitoring (“to keep a search on track”). Likewise, O’Day & Jeffries [11] examined the use of information search results by clients of professional information intermediaries and identified three categories of behaviour, including monitoring a known topic or set of variables over time and exploring a topic in an undirected fashion. They also observed that a given search scenario would often evolve into a series of interconnected searches, delimited by triggers and stop conditions that signalled transitions between modes within an overall scenario.

Cool & Belkin [3] proposed a classification of interaction with information which included evaluate and comprehend. They also proposed create and modify, which together reflect aspects of our synthesize mode.

Ellis and his colleagues [4, 5, 6] developed a model consisting of a number of broad information seeking behaviours, including monitoring and verifying(“checking the information and sources found for accuracy and errors”). In addition, his browsing mode (“semi-directed searching in an area of potential interest”) aligns with our definition of explore. He also noted that it is possible to display more than one behaviour at any given time. In revisiting Ellis’s findings among social scientists, Meho and Tibbo [10] identified analysing (although they did not elaborate on it in detail). More recently, Makri et al [8] proposed searching(“formulating a query in order to locate information”), which reflects to our own definition of locate.

In addition to the research-oriented models outlined above, we should also consider practitioner-oriented frameworks. Spencer [14] suggests four modes of information seeking, including known-item (a subset of our locate mode) andexploratory (which mirrors our definition of explore). Lamantia [7] also identifies four modes, including monitoring.

In this paper, we use the characteristics of the models above as a lens to interpret the behaviours expressed in a new source of empirical data. We also examine the combinatorial nature of the modes, extending Ellis’s [5] concept of mode co-occurrence to identify and define common patterns and sequences of information seeking behaviour.

3 Studying Search Behaviour

3.1 Data Acquisition

The primary source of data in this study is a set of 381 information needs captured during client engagements involving the development of a number of custom search applications. These information needs take the form of ‘micro-scenarios’, i.e. a brief narrative that illustrates the end user’s goal and the primary task or action they take to achieve it, for example:

• Find best offers before the others do so I can have a high margin.
• Get help and guidance on how to sell my car safely so that I can achieve a good price.
• Understand what is selling by area/region so I can source the correct stock.
• Understand a portfolio’s exposures to assess investment mix
• Understand the performance of a part in the field so that I can determine if I should replace it

The scenarios were collected as part of a series of requirements workshops involving stakeholders and customer-facing staff from various client organisations. A proportion of these engagements focused on consumer-oriented site search applications (resulting in 277 scenarios) and the remainder on enterprise search applications (104 scenarios).

The scenarios were generated by participants in breakout sessions and subsequently moderated by the workshop facilitator in a group session to maximise consistency and minimise redundancy or ambiguity. They were also prioritised by the group to identify those that represented the highest value both to the end user and to the client organisation.

This data possesses a number of unique properties. In previous studies of information seeking behaviour (e.g. [5], [10]), the primary source of data has traditionally been interview transcripts that provide an indirect, verbal account of end user information behaviours.  By contrast, the current data source represents a self-reported account of information needs, generated directly by end users (although a proportion were captured via proxy, e.g. through customer facing staff speaking on behalf of the end users). This change of perspective means that instead of using information behaviours to infer information needs and design insights, we can adopt the converse approach and use the stated needs to infer information behaviours and the interactions required to support them.

Moreover, the scope and focus of these scenarios represents a further point of differentiation. In previous studies, (e.g. [8]), measures have been taken to address the limitations of using interview data by combining it with direct observation of information seeking behaviour in naturalistic settings. However, the behaviours that this approach reveals are still bounded by the functionality currently offered by existing systems and working practices, and as such do not reflect the full range of aspirational or unmet user needs encompassed by the data in this study.

Finally, the data is unique in that is constitutes a genuine practitioner-oriented deliverable, generated expressly for the purpose of designing and delivering commercial search applications. As such, it reflects a degree of realism and authenticity that interview data or other research-based interventions might struggle to replicate.

3.2 Data Analysis

These scenarios were manually analyzed to identify themes or modes that appeared consistently throughout the set, using a number of iterations of a ‘propose-classify-refine’ cycle based on that of Rose & Levinson [14]. Inevitably, this process was somewhat subjective, echoing the observations made by Bates [1] in her work on search tactics:

“While our goal over the long term may be a parsimonious few, highly effective tactics, our goal in the short term should be to uncover as many as we can, as being of potential assistance. Then we can test the tactics and select the good ones. If we go for closure too soon, i.e., seek that parsimonious few prematurely, then we may miss some valuable tactics.”

In this respect, the process was partially deductive, in applying the insights from existing models to classify the data in a top-down manner. But it was also partially inductive, applying a bottom-up, grounded analysis to identify new types of behaviour not present in the original models or to suggest revised definitions of existing behaviours.

A number of the scenarios focused on needs that did not involve any explicit information seeking or use behaviour, e.g. “Achieve a good price for my current car”. These were excluded from the analysis. A further number were incomplete or ambiguous, or were essentially feature requests (e.g. “Have flexible navigation within the page”), and were also excluded.

The process resulted in the identification of nine primary search modes, which are defined below along with an example scenario (from the domain of consumer-oriented search):

1. Locate: To find a specific (possibly known) item, e.g. “Find my reading list items quickly”. This mode encapsulates the stereotypical ‘findability’ task that is so commonly associated with site search. It is consistent with (but a superset of) Spencer’s [14] known item search mode. This was the most frequent mode in the site search scenarios (120 instances, which contrasts with just 2 for enterprise search).

2. Verify: To confirm that an item meets some specific, objective criterion, e.g. “See the correct price for singles and deals”. Often found in combination with locating, this mode is concerned with validating the accuracy of some data item, comparable to that proposed by Ellis et al.  [5] (39 site search instances, 4 for enterprise search).

3. Monitor: Maintain awareness of the status of an item for purposes of management or control, e.g. “Alert me to new resources in my area”. This activity focuses on the state of asynchronous responsiveness and is consistent with that of Bates [1], O’Day and Jeffries [11], Ellis [4], and Lamantia [7] (13 site search instances, 17 for enterprise search).

4. Compare: To identify similarities & differences within a set of items, e.g. “Compare cars that are my possible candidates in detail”. This mode has not featured prominently in most of the previous models (with the possible exception of Marchionini’s), but accounted for a significant proportion of enterprise search behaviour [13]. Although a common feature on many ecommerce sites, it occurred relatively infrequently in the site search data (2 site search instances, 16 for enterprise search).

5. Comprehend: To generate independent insight by interpreting patterns within a data set, e.g. “Understand what my competitors are selling”. This activity focuses on the creation of knowledge or understanding and is consistent with that of Cool & Belkin [3] and Marchionini [9] (50 site search instances, 12 for enterprise search).

6. Evaluate: To use judgement to determine the value of an item with respect to a specific goal, e.g. “I want to know whether my agency is delivering best value”. This mode is similar in spirit to verify, in that it is concerned with validation of the data. However, while verify focuses on simple, objective fact checking, our conception of evaluate involves more subjective, knowledge-based judgement, similar to that proposed by Cool & Belkin [3] (61 site search instances, 78 for enterprise search).

7. Explore: To investigate an item or data set for the purpose of knowledge discovery, e.g. “Find useful stuff on my subject topic”. In some ways the boundaries of this mode are less prescribed than the others, but what the instances share is the characteristic of open ended, opportunistic search and browsing in the spirit of O’Day and Jeffries [11] exploring a topic in an undirected fashion and Spencer’s [14] exploratory (110 site search instances, 16 for enterprise search).

8. Analyze: To examine an item or data set to identify patterns & relationships,e.g. Analyze the market so I know where my strengths and weaknesses are”. This mode features less prominently in previous models, appearing as a sub-component of the processing stage in Meho & Tibbo’s [10] model, and overlapping somewhat with Cool & Belkin’s [3] organize. This definition is also consistent with that of Makri et al. [8], who identified analysing as an important aspect of lawyers’ interactive information behaviour and defined it as “examining in detail the elements or structure of the content found during information-seeking.” (p. 630). This was the most common element of the enterprise search scenarios (58 site search instances, 84 for enterprise search).

9. Synthesize: To create a novel or composite artefact from diverse inputs, e.g. “I need to create a reading list on celebrity sponsorship”. This mode also appears as a sub-component of the processing stage in Meho & Tibbo’s [10] model, and involves elements of Cool & Belkin’s [3] create and use. Of all the modes, this one is the most commonly associated with information use in its broadest sense (as opposed to information seeking). It was relatively rare within site search (5 site search instances, 15 for enterprise search).

Although the modes were generated from an independent data source and analysis process, we have retrospectively explored the degree to which they align with existing frameworks, e.g. Marchionini’s [8]. In this context, locate, verify, andmonitor could be described as lower-level ‘lookup’ modes, compare, comprehend, and evaluate as ‘learn’ modes and explore, analyze, and synthesize as higher-level ‘investigate’ modes.

4 Mode Sequences and Patterns

The modes defined above provide an insight into the needs of users of site search and enterprise search applications and a framework for understanding human information seeking behaviour. But their real value lies not so much in their occurrence as individual instances but in the patterns of co-occurrence they reveal. In most scenarios, modes combine to form distinct chains and patterns, echoing the transitions observed by O’Day and Jeffries [11] and the combinatorial behaviour alluded to by Ellis [5], who suggested that information behaviours can often be nested or displayed in parallel.

Typically these patterns consist of chains of length two or three, often with one particular mode playing a dominant role. Site search, for example, was characterized by the following patterns:

By contrast, enterprise search was characterized by a larger number of more diverse sequences, such as:

A further insight into these patterns can be obtained by presenting them in diagrammatic form. Figure 1 illustrates sequences 1-3 above plus other commonly found site search patterns as a network (with sequence numbers shown on the arrows). It shows how certain modes tend to function as “terminal” nodes, i.e. entry points or exit points for a given scenario. For example, Explore typically functions as an opening, while Comprehend and Evaluate function in closing a scenario. Analyze typically appears as a bridge between an opening and closing mode. The shading indicates the mode ‘level’ alluded to earlier: light tones indicate ‘lookup’ modes, mid tones are the ‘learn’ modes, and dark tones are the ‘investigate’ modes.

Fig. 1. Mode network for site search

Figure 2 illustrates sequences 4-8 above plus other commonly found patterns in the enterprise search data.

Fig. 2. Mode network for enterprise search

The patterns described above allow us to reflect on some of the differences between the needs of site search users and those of enterprise search. Site search, for example, is characterized by an emphasis on simpler “lookup” behaviours such as Locate and Verify (120 and 39 instances respectively); modes which were relatively rare in enterprise search (2 and 4 instances respectively). By contrast, enterprise search is characterized by higher-level “learn” and “investigate” behaviours such as Analyze and Evaluate (84 and 78 instances respectively, compared to 58 and 61 for site search). Interestingly, in neither case was the stereotype of ‘search equals findability’ borne out: even in site search (whereLocate was the most common mode), known-item search was accountable for no more than a quarter of all instances.

But perhaps the biggest difference is in the composition of the chains: enterprise search is characterised by a wide variety of heterogeneous chains, while site searched focuses on a small number of common trigrams and bigrams. Moreover, the enterprise search chains often displayed a fractal nature, in which certain chains were embedded within or triggered by others, to create larger, more complex sequences of behaviour.

5 Design Implications

Although the model offers a useful framework for understanding human information seeking behaviour, its real value lies in its use as a practical design resource. As such, it can provide guidance on issues such as:

• the features and functionality that should be available at specific points within a system;
• the interaction design of individual functions or components;
• the design cues used to guide users toward specific areas of task interface.

Moreover, the model also has significant implications for the broader aspects of user experience design, such as the alignment between the overall structure or concept model of a system and its users’ mental models, and the task workflows for various users and contexts. This broader perspective addresses architectural questions such as the nature of the workspaces required by a given application, or the paths that users will take when navigating within a system’s structure.  In this way, the modes also act as a generative tool for larger, composite design issues and structures.

5.1 Individual modes

On their own, each of the modes describes a type of behaviour that may need to be supported by a given information system’s design. For example, an online retail site should support locating and comparing specific products, and ideally alsocomprehending differences and evaluating tradeoffs between them. Likewise, an enterprise application for electronic component selection should support monitoringand verifying the suitability of particular parts, and ideally also analyzing andcomprehending any relevant patterns and trends in their lifecycle. By understanding the anticipated search modes for a given system, we can optimize the design to support specific user behaviours. In the following section we consider individual instances of search modes and explore some of their design implications.

Locate

This mode encapsulates the stereotypical ‘findability’ task that is so commonly associated with site search. But support for this mode can go far beyond simple keyword entry. For example, by allowing the user to choose from a list of candidates, auto-complete transforms the query formulation problem from one of recall into one of recognition (Figure 3).

Fig. 3. Auto-complete supports locating

Likewise, Amazon’s partial match strategy deals with potentially failed queries by identifying the keyword permutations that are likely to produce useful results. Moreover, by rendering the non-matching keywords in strikethrough text, it facilitates a more informed approach to query reformulation (Figure 4).

Fig 4: Partial matches support Locating

Verify

In this mode, the user is inspecting a particular item and wishing to confirm that it meets some specific criterion.  Google’s image results page provides a good example of this (see Figure 5).

Fig 5: Search result previews support verification

On mouseover, the image is zoomed in to show a magnified version along with key metadata, such as filename, image size, caption, and source. This allows the user to verify the suitability of a specific result in the context of its alternatives. Likewise, there may be cases where the user needs to verify a particular query rather than a particular result. In providing real-time feedback after every key press, Google Instant supports verification by previewing the results that will be returned for a given query (Figure 6). If the results seem unexpected, the user can check the query for errors or try alternative spellings or keyword combinations.

Fig 6: Instant results supports verification of queries

Compare

The Compare mode is fundamental to online retail, where users need to identify the best option from the choices available. A common technique is to provide a custom view in which details of each item are shown in separate columns, enabling rapid comparison of product attributes. Best Buy, for example, supports comparison by organising the attributes into logical groups and automatically highlighting the differences (Figure 7).

Fig 7: Separate views support product comparison

But comparison is not restricted to qualitative attributes. In financial services, for example, it is vital to compare stock performance and other financial instruments with industry benchmarks. Google Finance supports the comparison of securities through a common charting component (Figure 8).

Fig 8: Common charts allow comparison of quantitative data

Explore

A key principle in exploring is differentiating between where you are going andwhere you have already been. In fact, this distinction is so important that it has been woven into the fabric of the web itself; with unexplored hyperlinks rendered in blue by default, and visited hyperlinks shown in magenta. Amazon takes this principle a step further, through components such as a  ‘Recent Searches’ panel showing the previous queries issued in the current session, and a ‘Recent History’ panel showing the items recently viewed (Figure 9).

Fig 9: Recent history supports exploration

Another simple technique for encouraging exploration is through the use of “see also” panels. Online retailers commonly use these to promote related products such as accessories and other items to complement an intended purchase. An example of this can be seen at Food Network, in which featured videos and products are shown alongside the primary search results (Figure 10).

Fig 10: ‘See Also’ panels support exploration

A further technique for supporting exploration is through the use of auto-suggest. While auto-complete helps users get an idea out of their heads and into the search box, auto-suggest throws new ideas into the mix. In this respect, it helps users explore by formulating more useful queries than they might otherwise have thought of on their own. Home Depot, for example, provides a particularly extensive auto-suggest function consisting of product categories, buying guides, project guides and more, encouraging the discovery of new product ideas and content (Figure 11).

Fig 11: Auto-suggest supports exploratory search

Analyze

In modes such as exploring, the user’s primary concern is in understanding theoverall information space and identifying areas to analyze in further detail. Analysis, in this sense, goes hand in hand with exploring, as together they present complementary modes that allow search to progress beyond the traditional confines of information retrieval or ‘findability’.

A simple example of this could be found at Google patents (Figure 12). The alternate views (Cover View and List View) allow the user to switch between rapid exploration (scanning titles, browsing thumbnails, looking for information scent) and a more detailed analysis of each record and its metadata.

Fig 12: Alternate views support mode switching between exploration and analysis

In the above example the analysis focuses on qualitative information derived from predominantly textual sources. Other applications focus on quantitative data in the form of aggregate patterns across collections of records. NewsSift, for example, provided a set of data visualizations which allowed the user to analyze results for a given news topic at the aggregate level, gaining an insight that could not be obtained from examining individual records alone (Figure 13).

Fig 13: Visualizations support analysis of quantitative information

5.2 Composite patterns

The examples above represent instances of individual modes, showing various ways they can be supported by one or more aspects of a system’s design. However, a key feature of the model is its emphasis on the combinatorial nature of modes and the patterns of co-occurrence this reveals [12]. In this respect, its true value is in helping designers to address more holistic, larger scale concerns such as the appropriate structure, concept model, and organizing principles of a system, as well as the functional and informational content of its major components and connections between them.

Design at this level relies on translating composite modes and chains that represent sense-making activities – often articulated as user journeys through a task and information space – into interaction components that represent meaningful combinations of information and discovery capabilities [13].  These components serve as ‘building blocks’ that designers can assemble into larger composite structures to create a user experience that supports the anticipated user journeys and aligns with their users’ mental models [14].

The popular micro-blogging service twitter.com provides a number of examples of the correspondence between composite modes and interaction components assembled at various levels to provide a coherent user experience architecture.

Header Bar

The header bar at the top of most pages of twitter.com combines several informational and functional elements together in a single component that supports a number of modes and mode chains (Figure 14). It includes four dynamic status indicators that address key aspects of twitter’s concept model and the users’ mental models:

• the presence of new tweets by people the user follows
• interactions with other twitter users such as following them or mentioning them in a tweet
• activity related to the user’s profile, such as their latest tweets and shared media
• people, topics, or items of interest suggested by the systems recommender functions

These status indicator icons update automatically and provide links to specific pages in the twitter.com application architecture that provide further detail on each area of focus. The header bar thus enables Monitoring of a user’s activity within the full scope of the twitter.com network; i.e. its content, members, their activities, etc.  The header bar also enables Monitoring activity within almost all the workspaces that users encounter in the course of their primary journeys throughtwitter.com.

Fig. 14. twitter.com Header Bar

The Strategic Oversight chain (Monitor – Analyze – Evaluate) is a fundamental sequence for twitter users, repeated frequently with different aspects of the user’s profile. The header bar supports the first step of this chain, in which users Monitor the network for content and activity of interest to them, and then transition to Analysis and Evaluation of that activity by navigating to destination pages for further detail.

The header bar also includes a search box featuring auto-complete and auto-suggest functionality, which provides support for the Qualified Search mode chain (Locate – Verify). The search box also enables users to initiate many other mode chains by supporting the Explore mode. These include Exploratory Search (Explore – Analyze – Evaluate), Insight-driven Search (Explore – Analyze – Comprehend), and Opportunity-driven Search (Explore – Locate – Evaluate). All these mode chains overlap by sharing a common starting point. This is one of the most readily recognizable kinds of composition, and often corresponds to a single instance of a particular interaction component.

The header bar includes support for posting or Synthesizing new tweets, reflecting the fact that the creation of new content is probably the second most important individual mode (after Monitoring). A menu of links to administrative pages and functions for managing one’s twitter account completes the content of the header bar.

Individual Tweets

The individual tweets and activity updates that make up the stream at the heart of the primary workspace are the most important interaction components of the twitter experience, and their design shows a direct correspondence to many composite modes and chains (Figure 15). Individual items provide the content of a tweet along with the author’s public name, their twitter username, profile image, and the time elapsed since the tweet’s creation. Together, these details allow users to Compare and Comprehend the content and significance of tweets in their own stream.  As users read more tweets and begin to recognize authors and topics, they can Compare, Analyze, and Evaluate them.  The indicators of origin and activity allow users to Compare and Comprehend the topics and interests of other twitter users.

Fig. 15. Individual Tweet

Options to invoke a number of functions that correspond to other discovery modes are embedded within the individual items in the stream. For example, if an update was retweeted, it is marked as such with the original author indicated and their profile page linked. It also shows the number of times the tweet has been retweeted and favorited, with links that open modal previews of the list of users who did so. This supports Monitoring, Exploration and Comprehension of the significance and attention an individual tweet has received, while the links support Location, Verification and Monitoring of the other users who retweeted or favorited it.

Public profile names and usernames are linked to pages which summarize the activities and relationships of the author of a tweet, enabling users to Locate and Verify authors, then transition to Monitoring, Exploring and Comprehending their activities, interests, and how they are connected to the rest of the twitter network.

Hashtags are presented with distinct visual treatment.  When users click on one, it initiates a search using the hashtag, allowing users to Locate, Explore, Comprehend, and Analyze the topic referred to, any conversations in which the tag is mentioned, and the users who employ the tag.

Fig. 16. Expanded Tweet

Longer tweets are truncated, offering an ‘Expand’ link which opens a panel displaying the number of retweets and favourites and the images of the users who did so, along with the date and time of authoring and a link to a ‘details’ page for a permanent URL that other users and external services can reference (Figure 16). This sort of truncation enables users to more easily Explore the full set of tweets in a stream and Locate individual items of interest. Conversely, the ‘Expand’ panel allows the user to more easily Explore and Comprehend individual items.

Tweets that contain links to other tweets offer a ‘View tweet’ link, which opens a panel displaying the full contents of the original tweet, the date and time of posting, the number of retweets and favorites and a preview list of the users who did so.  The ‘View tweet’ link thus supports the Locate, Explore, and Comprehend modes for individual updates.

Tweets that contain links to digital assets such as photos, videos, songs, presentations, and documents, offer users the ability to preview these assets directly within an expanded display panel, providing support for the Locate, Explore, and Comprehend modes. These previews link to the source of the assets, enabling users to Locate them.  Users can also ‘flag’ media for review by twitter (e.g. due to violation of policies about sensitive or illegal imagery) – which is a very specific form of Evaluation.

Fig. 17. Tweet Displaying a Photo

Tweets that contain links to items such as articles published by newspapers, magazines, and journals, or recognized destinations such as Foursquare and Google + pages, offer a ‘Summary’ link (Figure 17). This link opens a panel that presents the first paragraph of the article or destination URL, an image from the original publisher, and a list of users who have retweeted or favorited it, thus supporting Location, Exploration and Verification of the linked item.

A text input field seeded with the author’s username allows users to reply to specific tweets directly from an individual update. Users can also ‘retweet’ items directly from the list. Both functions are forms of Synthesis, and encourage users to create further content and relationships within the network.

Users can mark tweets as ‘favorites’ to indicate the importance or value of these tweets to others; a clear example of the Evaluation mode. Favorites also allow users to build a collection of tweets curated for retrieval and interpretation, enabling the Locate, Compare, Comprehend, and Analyze modes for tweets as individual items or as groups.

A ‘More’ link opens a menu offering ‘Email Tweet’ and ‘Embed Tweet’ options, allowing users to initiate tasks that take tweets outside the twitter environment.  These two functions support information usage modes, rather than search anddiscovery modes, so their distinct treatment – invoked via a different interaction than the other functions – is consistent with the great emphasis the twitter experience places on discovery and sense making activities.

If the tweet is part of a conversation, a ‘View this conversation’ link allows readers to open a panel that presents related tweets and user activity as a single thread, accompanied by a reply field.  This provides support for the Locate, Explore, Comprehend, Analyze, Evaluate and Synthesize modes (Figure 18).

Fig. 18. Tweet Showing a Conversation

The informational and functional content presented by individual items in their various forms enables a number of mode chains. These include Strategic Oversight, in which users maintain awareness of conversations, topics, other users, and activities; Strategic Insight, wherein users focus on and derive insight into conversations, topics, and other users; and Comparative Synthesis, in which users realize new insights and create new content through direct engagement with conversations, topics, and other users.

In a manner similar to the search box, this interaction component serves as an initiation point for a number of mode chains, including Exploratory Search, Insight-driven Search, and Opportunity-driven Search. Individual tweets thus combine support for many important modes and mode chains into a single interaction component.  As a consequence, they need to be relatively rich and ‘dense’, compacting much functionality into a single interaction component, but this reflects their crucial role in the user journeys that characterize the twitter experience.

Primary Workspaces and Pages

In the previous section we reviewed the correspondence between groups of modes and the interaction components of a user experience. In this section, we review the ways in which modes and chains impact the composition and presentation of the next level of UX structure within the system: work spaces.

The primary workspaces of twitter.com all emphasize interaction with a stream of individual updates, but the focus and content vary depending on the context. On the Home page, for example, the central stream consists of tweets from people the user follows, while on the ‘Me’ page the stream consists of the tweets created by the user (Figure 19). However, the layout of these pages remains consistent: the workspace is dominated by a single central stream of individual updates. The primary interaction mode for this stream is Monitoring, evident from the count of new items added to the network since the last page refresh.

Fig. 19. twitter.com Home Workspace

The placement of the header bar at the top of all of the primary workspaces is a design decision that reflects the primacy of Monitoring as a mode of engagement with the twitter service; supporting its role as a persistent ‘background’ mode of discovery independent of the user’s current point in a task or journey, and its role as a common entry point to the other mode chains and user journeys.

The consistent placement of the ‘Compose new Tweet’ control in upper right corner of the workspace reflects known interaction design principles (corners are the second most easily engaged areas of a screen, after the centre) and the understanding that Synthesis is the second most important single mode for the twitter service.

The content of the individual updates attracts and retains users’ attention very effectively: the majority of the actions a user may want to take in regard to a tweet (or any of the related constructs in twitter’s concept model such as conversations, hash tags, profiles, linked media, etc.) are directly available from the interaction component.  In some cases, these actions are presented via modal or lightbox preview, wherein the user’s focus is ‘forced’ onto a single element – thus maintaining the primacy of the stream.  In others, links lead to destination pages that switch the user’s focus to a different subject – another user’s profile, for example – but in most of these cases the structure of the workspace remains consistent: a two column body surmounted by the ubiquitous header bar. There is little need to look elsewhere in the workspace, unless the user needs to check the status of one of the broader aspects of their account, at which point the header bar provides appropriate functionality as discussed above.

The absence of a page footer – scrolling is ‘infinite’ on the primary pages oftwitter.com – reflects the conscious decision to convey updates as an endless, dynamic stream.  This encourages users to continue scrolling, increasing Exploration activity, and enhancing users’ Comprehension of additional updates – which benefits twitter’s business by increasing the attention users direct toward the service.

Although the two-tier, stream-centred structure of twitter’s primary workspaces remains consistent, there are variations in the composition of the left column (Figure 20). On the Home page, for example, the left column offers four separate components. The first is a summary of the user’s profile, including a profile image, a link to their profile page, counts of their tweets, followers, and the people they follow, and a ‘compose new tweet’ box.  This is another example of a component supporting a composite of modes.

Fig. 20. Twitter Home Page – Left Column

The core purpose is to enable users to Monitor the most important aspects of their own account via the counts.  The links provide direct Locate functionality for followers, tweets, and accounts the user follows; and also serve as a point of departure for the same mode chains that can be initiated from the header bar.  The ‘compose new tweet’ function encourages users to create updates, underlining the importance of Synthesis as the source of new content within the twitter network.

User Experience Architecture

The twitter.com experience is intended to support a set of user journeys consisting largely of search and discovery tasks which correspond with specific monitoring and search-related mode chains. Further, we can see that patterns of recurrence, intersection, overlap, and sequencing in the aggregate set of search and discovery modes are substantially reflected in twitter’s user experience architecture.

From a structural design perspective, the core [16] of the twitter.com user experience architecture is a set of four interaction consoles, each of which focuses on monitoring a distinct stream of updates around the most important facets of thetwitter.com concept model: the content and activities of people in the user’s personal network (Home); interactions with other users (Interactions); the user’s profile (@Me); and a digest of content from all users in the twitter.com network (Discover) (Figure 21).

The core monitoring consoles are supported by screens that assist and encourage users to expand their personal networks through location and exploration tools; these include ‘Find friends’, ‘Who to follow’ ‘Browse categories’, and the search results page.

Fig. 21. Twitter.com Discover Workspace

Specific landing pages provide monitoring and curation tools for the different types of relationships users can establish in the social graph: follow and un-follow, followers and following, public and private accounts, list memberships, etc.  A small set of screens provides functionality for administering the user’s account, such as ‘Settings’.

Underlying this user experience architecture is a concept model consisting primarily of a small set of social objects – tweets, conversations, profiles, shared digital assets, and lists thereof – linked together by search and discovery verbs. A relatively simple information architecture establishes the set of categories used to identify these objects by topic, similarity, and content (Figure 22).

In its holistic and granular aspects, the twitter user experience architecture aligns well with users’ mental models for building a profile and participating in an ongoing stream of conversations. However, what emerges quite quickly from analysis of the twitter concept model and user experience architecture is the role of search and discovery modes in both atomic and composite forms at every level of twitter’s design. Rather than merely subsuming modes as part of some larger activity, many of the most common actions users can take with twitter’s core interaction objects correspond directly to modes themselves.

Fig. 22. Twitter.com User Experience Architecture

The individual tweet component is a prime example: the summaries of author profiles and their recent activity are a composite of the Locate, Explore and Comprehend modes (Figure 23). Evidently, the presentation, labelling, and interaction design may reflect adaptations specific to the language and mental model of the twitter environment, but the activities are clearly recognizable. The ‘Show conversation’ function discussed above also reflects direct support to Locate, Explore and Comprehend a conversation object as a single interaction.

Fig. 23. Twitter Profile Summary

Because the twitter.com experience is so strongly centred on sense-making, search and discovery modes often directly constitute the activity paths connecting one object to another within the user experience architecture.  In this sense, the modes and chains could be said to act as a ‘skeleton’ for twitter.com, and are directly visible to an unprecedented degree in the interaction design built on that skeleton.

6 Discussion

The model described in this paper encompasses a range of information seeking behaviours, from elementary lookup tasks through to more complex problem-solving activities. However, the model could also be framed as part of a broader set of information behaviours, extending from ‘acquisition’ oriented tasks at one end of the spectrum to ‘usage’ oriented activities at the other (Figure 24). In this context, modes can span more than one phase. For example, Explore entails a degree ofinteraction coupled with the anticipation of further discovery, i.e. acquisition.  Likewise, Evaluate implies a degree of interaction in the pursuit of some higher goal or purpose to which the output will be put, i.e. usage.

It would appear that with the possible exception of synthesize, there are no exclusively usage-oriented behaviours in the model. This may suggest that the model is in some senses incomplete, or may simply reflect the context in which the data was acquired and the IR-centric processes by which it was analysed.

Reducing the ‘scope’ of the model such that modes serve only as descriptors of distilled sense-making activity independent of context (such as the user’s overall goal and the nature of the information assets involved) may help clarify the relationship between acquisition, interaction and usage phases. In this perspective, there appears to be a form of ‘parallelism’ in effect; with users simultaneously undertaking activities focused on an overall goal, such as Evaluating the quality of a financial instrument, while also performing activities focused on narrower information-centred objectives such as Locating and Verifying the utility of the information assets necessary for them to complete the Evaluation.  These ‘parallel’ sets of activities – one focused on information assets in service to a larger goal, and the other focused on the goal itself – can be usefully described in terms of modes, and what is more important, seem intertwined in the minds of users as they articulate their discovery needs.

Fig. 24. From information acquisition to information use

A key feature of the current model is its emphasis on the combinatorial nature of search modes, and the value this offers as a framework for expressing complex patterns of behaviour. Evidently, such an approach is not unique: Makri (2008), for example, has also previously explored the concept of mode chains to describe information seeking behaviours observed in naturalistic settings. However, his approach was based on the analysis of complex tasks observed in real time, and as such was less effective in revealing consistent patterns of atomic behaviour such as those found in the current study.

Conversely, this virtue can also be a shortcoming: the fact that simple repeating patterns can be extracted from the data may be as much an artefact of the medium as it is of the information needs it contains. These scenarios were expressly designed to be a concise, self-contained deliverable in their own right, and applied as a simple but effective tool in the planning and prioritisation of software development activities. This places a limit on the length and sophistication of the information needs they encapsulate, and a natural boundary on the scope and extent of the patterns they represent. Their format also allows a researcher to apply perhaps an unrealistic degree of top-down judgement and iteration in aligning the relative granularity of the information needs to existing modes; a benefit that is less readily available to those whose approach involves real-time, observational data.

A further caveat is that in order to progress from understanding an information need to identifying the information behaviours required to satisfy those needs, it is necessary to speculate on the behaviours that a user might perform when undertaking a task to satisfy the need. It may transpire that users actually perform different behaviours which achieve the same end, or perform the expected behaviour but through a combination of other nested behaviours, or may simply satisfy the need in a way that had not been envisaged at all.

Evidently, the process of inferring information behaviour from self-reported needs can never be wholly deterministic, regardless of the consistency measures discussed in Section 3.1. In this respect, further steps should be taken to operationalize the process and develop some independent measure of stability or objectivity in its usage, so that its value and insights can extend reliably to the wider research community.

The compositional behaviour of the modes suggests further open questions and avenues for research. One of these is the nature of compositionality itself: one the one hand it could be thought of as a pseudo-linguistic grammar, with bigrams and trigrams of modes that combine in turn to form larger sequences, analogous to coherent “sentences”. In this context, the modes act as verbs, while the associated objects (users, information assets, processes etc.) become the nouns. The occurrence of distinct ‘opening’ and ‘closing’ modes in the scenarios would seem to further support this view. However, in some scenarios the transitions between the modes are far less apparent, and instead they could be seen as applying in parallel, like notes combining in harmony to form a musical chord. In both cases, the degree and nature of any such compositional rules needs further empirical investigation. This may reveal other dependencies yet to be observed, such as the possibility alluded to earlier of higher-level behaviours requiring the completion of certain lower level modes before they themselves can terminate.

The process of mapping from modes to design interventions also reveals further observations on the utility of information models in general. Despite their evident value as analytical frameworks and their popularity among researchers (Bates’ Berrypicking model has been cited over 1,000 times, for example), few have gained significant traction within the design community, and fewer still are adopted as part of the mainstream working practices of system design practitioners.

In part, this may be simply a reflection of imperfect channels of communication between the research and design communities. However, it may also reflect a growing conceptual gap between research insights on the one hand and corresponding design interventions on the other. It is likely that the most valuable theoretical models will need to strike a balance between flexibility (the ability to address a variety of domains and problems), generative power (the ability to express complex patterns of behaviour) and an appropriate level of abstraction (such that design insights are readily available; or may be inferred with minimal speculation).

7 Conclusions

In this paper, we have examined the needs and behaviours of individuals across a wide range of search and discovery scenarios. We have proposed a model of information seeking behaviour which has at its core a set of modes that people regularly employ to satisfy their information needs. In so doing, we explored a novel, goal-driven approach to eliciting user needs, and identified some key differences in user behaviour between site search and enterprise search.

In addition, we have demonstrated the value of the model as a framework for expressing complex patterns of search behaviour, extending the IR concept of information-seeking to embrace a broader range of information interaction and use behaviours. We propose that our approach can be adopted by other researchers who want to adopt a ‘needs first’ perspective to understanding information behaviour.

By illustrating ways in which individual modes are supported in existing search applications, we have made a practical contribution that helps bridge the gap between investigating search behaviour and designing applications to support such behaviour. In particular, we have demonstrated how modes can serve as an effective design tool across varied levels of system design: concept model, UX architecture, interaction design, and visual design.

References

 

I’ve posted a substantial readout summarizing some of the more salient findings from a long-running programmatic research program into data science. This deck shares synthesized findings around many of the facets of data science as a discipline, including practices, workflow, tools, org models, skills, etc. This readout distills a very wide range of inputs, including; direct interviews, field-based ethnography, community participation (real-world and on-line), secondary research from industry and academic sources, analysis of hiring and investment activity in data science over several years, descriptive and definitional artifacts authored by practitioners / analysts / educators, and other external actors, media coverage of data science, historical antecedents, the structure and evolution of professional disciplines, and even more.

I consider it a sort of business-anthropology-style investigation of data science, conducted from the viewpoint of product making’s primary aspects; strategy, management, design, and delivery.

I learned a great deal during the course of this effort, and expect to continue to learn, as data science will continue to evolve rapidly for the next several years.

Data science practitioners looking at this material are invited to provide feedback about where these materials are accurate or inaccurate, and most especially about what is missing, and what is coming next for this very exciting field.

One of the essential patterns of science and industry in the modern era is that new methods for understanding — what I’ll call sensemaking from now on — often emerge hand in hand with new professional and scientific disciplines.  This linkage between new disciplines and new methods follows from the  deceptively simple imperative to realize new types of insight, which often means analysis of new kinds of data, using new techniques, applied from newly defined perspectives. New viewpoints and new ways of understanding are literally bound together in a sort of symbiosis.

One familiar example of this dynamic is the rapid development of statistics during the 18th and 19th centuries, in close parallel with the rise of new social science disciplines including economics (originally political economy) and sociology, and natural sciences such as astronomy and physics.  On a very broad scale, we can see the pattern in the tandem evolution of the scientific method for sensemaking, and the codification of modern scientific disciplines based on precursor fields such as natural history and natural philosophy during the scientific revolution.

Today, we can see this pattern clearly in the simultaneous emergence of Data Science as a new and distinct discipline accompanied by Empirical Discovery, the new sensemaking and analysis method Data Science is pioneering.  Given its dramatic rise to prominence recently, declaring Data Science a new professional discipline should inspire little controversy. Declaring Empirical Discovery a new method may seem bolder, but when we with the essential pattern of new disciplines appearing in tandem with new sensemaking methods in mind, it is more controversial to suggest Data Science is a new discipline that lacks a corresponding new method for sensemaking.  (I would argue it is the method that makes the discipline, not the other way around, but that is a topic for fuller treatment elsewhere)

What is empirical discovery?  While empirical discovery is a new sensemaking method, we can build on two existing foundations to understand its distinguishing characteristics, and help craft an initial definition.  The first of these is an understanding of the empirical method. Consider the following description:

“The empirical method is not sharply defined and is often contrasted with the precision of the experimental method, where data are derived from the systematic manipulation of variables in an experiment.  …The empirical method is generally characterized by the collection of a large amount of data before much speculation as to their significance, or without much idea of what to expect, and is to be contrasted with more theoretical methods in which the collection of empirical data is guided largely by preliminary theoretical exploration of what to expect. The empirical method is necessary in entering hitherto completely unexplored fields, and becomes less purely empirical as the acquired mastery of the field increases. Successful use of an exclusively empirical method demands a higher degree of intuitive ability in the practitioner.”

Data Science as practiced is largely consistent with this picture.  Empirical prerogatives and understandings shape the procedural planning of Data Science efforts, rather than theoretical constructs.  Semi-formal approaches predominate over explicitly codified methods, signaling the importance of intuition.  Data scientists often work with data that is on-hand already from business activity, or data that is newly generated through normal business operations, rather than seeking to acquire wholly new data that is consistent with the design parameters and goals of formal experimental efforts.  Much of the sensemaking activity around data is explicitly exploratory (what I call the ‘panning for gold’ stage of evolution – more on this in subsequent postings), rather than systematic in the manipulation of known variables.  These exploratory techniques are used to address relatively new fields such as the Internet of Things, wearables, and large-scale social graphs and collective activity domains such as instrumented environments and the quantified self.  These new domains of application are not mature in analytical terms; analysts are still working to identify the most effective techniques for yielding insights from data within their bounds.

The second relevant perspective is our understanding of discovery as an activity that is distinct and recognizable in comparison to generalized analysis: from this, we can summarize as sensemaking intended to arrive at novel insights, through exploration and analysis of diverse and dynamic data in an iterative and evolving fashion.

Looking deeper, one specific characteristic of discovery as an activity is the absence of formally articulated statements of belief and expected outcomes at the beginning of most discovery efforts.  Another is the iterative nature of discovery efforts, which can change course in non-linear ways and even ‘backtrack’ on the way to arriving at insights: both the data and the techniques used to analyze data change during discovery efforts.  Formally defined experiments are much more clearly determined from the beginning, and their definition is less open to change during their course. A program of related experiments conducted over time may show iterative adaptation of goals, data and methods, but the individual experiments themselves are not malleable and dynamic in the fashion of discovery.  Discovery’s emphasis on novel insight as preferred outcome is another important characteristic; by contrast, formal experiments are repeatable and verifiable by definition, and the degree of repeatability is a criteria of well-designed experiments.  Discovery efforts often involve an intuitive shift in perspective that is recountable and retraceable in retrospect, but cannot be anticipated.

Building on these two foundations, we can define Empirical Discovery as a hybrid, purposeful, applied, augmented, iterative and serendipitous method for realizing novel insights for business, through analysis of large and diverse data sets.

Let’s look at these facets in more detail.

Empirical discovery primarily addresses the practical goals and audiences of business (or industry), rather than scientific, academic, or theoretical objectives.  This is tremendously important, since  the practical context impacts every aspect of Empirical Discovery.

‘Large and diverse data sets’ reflects the fact that Data Science practitioners engage with Big Data as we currently understand it; situations in which the confluence of data types and volumes exceeds the capabilities of business analytics to practically realize insights in terms of tools, infrastructure, practices, etc.

Empirical discovery uses a rapidly evolving hybridized toolkit, blending a wide range of general and advanced statistical techniques with sophisticated exploratory and analytical methods from a wide variety of sources that includes data mining, natural language processing, machine learning, neural networks, bayesian analysis, and emerging techniques such as topological data analysis and deep learning.

What’s most notable about this hybrid toolkit is that Empirical Discovery does not originate novel analysis techniques, it borrows tools from established disciplines such information retrieval, artificial intelligence, computer science, and the social sciences.  Many of the more specialized or apparently exotic techniques data science and empirical discovery rely on, such as support vector machines, deep learning, or measuring mutual information in data sets, have established histories of usage in academic or other industry settings, and have reached reasonable levels of maturity.  Empirical discovery’s hybrid toolkit is  transposed from one domain of application to another, rather than invented.

Empirical Discovery is an applied method in the same way Data Science is an applied discipline: it originates in and is adapted to business contexts, it focuses on arriving at useful insights to inform business activities, and it is not used to conduct basic research.  At this early stage of development, Empirical Discovery has no independent and articulated theoretical basis and does not (yet) advance a distinct body of knowledge based on theory or practice. All viable disciplines have a body of knowledge, whether formal or informal, and applied disciplines have only their cumulative body of knowledge to distinguish them, so I expect this to change.

Empirical discovery is not only applied, but explicitly purposeful in that it is always set in motion and directed by an agenda from a larger context, typically the specific business goals of the organization acting as a prime mover and funding data science positions and tools.  Data Science practitioners effect Empirical Discovery by making it happen on a daily basis – but wherever there is empirical discovery activity, there is sure to be intentionality from a business view.  For example, even in organizations with a formal hack time policy, our research suggests there is little or no completely undirected or self-directed empirical discovery activity, whether conducted by formally recognized Data Science practitioners, business analysts, or others.

One very important implication of the situational purposefulness of Empirical Discovery is that there is no direct imperative for generating a body of cumulative knowledge through original research: the insights that result from Empirical Discovery efforts are judged by their practical utility in an immediate context.  There is also no explicit scientific burden of proof or verifiability associated with Empirical Discovery within it’s primary context of application.  Many practitioners encourage some aspects of verifiability, for example, by annotating the various sources of data used for their efforts and the transformations involved in wrangling data on the road to insights or data products, but this is not a requirement of the method.  Another implication is that empirical discovery does not adhere to any explicit moral, ethical, or value-based missions that transcend working context.  While Data Scientists often interpret their role as transformative, this is in reference to business.  Data Science is not medicine, for example, with a Hippocratic oath.

Empirical Discovery is an augmented method in that it depends on computing and machine resources to increase human analytical capabilities: It is simply impractical for people to manually undertake many of the analytical techniques common to Data Science.  An important point to remember about augmented methods is that they are not automated; people remain necessary, and it is the combination of human and machine that is effective at yielding insights.  In the problem domain of discovery, the patterns of sensemaking activity leading to insight are intuitive, non-linear, and associative; activites with these characteristics are not fully automatable with current technology. And while many analytical techniques can be usefully automated within boundaries, these tasks typically make up just a portion of an complete discovery effort.  For example, using latent class analysis to explore a machine-sampled subset of a larger data corpus is task-specific automation complementing human perspective at particular points of the Empirical Discovery workflow.  This dependence on machine augmented analytical capability is recent within the history of analytical methods.  In most of the modern era — roughly the later 17th, 18th, 19th and early 20th centuries — the data employed in discovery efforts was manageable ‘by hand’, even when using the newest mathematical and analytical methods emerging at the time.  This remained true until the effective commercialization of machine computing ended the need for human computers as a recognized role in the middle of the 20th century.

The reality of most analytical efforts — even those with good initial definition — is that insights often emerge in response to and in tandem with changing and evolving questions which were not identified, or perhaps not even understood, at the outset.  During discovery efforts, analytical goals and techniques, as well as the data under consideration, often shift in unpredictable ways, making the path to insight dynamic and non-linear.  Further, the sources of and inspirations for insight are  difficult or impossible to identify both at the time and in retrospect. Empirical discovery addresses the complex and opaque nature of discovery with iteration and adaptation, which combine  to set the stage for serendipity.

With this initial definition of Empirical Discovery in hand, the natural question is what this means for Data Science and business analytics?  Three thigns stand out for me.  First, I think one of the central roles played by Data Science is in pioneering the application of existing analytical methods from specialized domains to serve general business goals and perspectives, seeking effective ways to work with the new types (graph, sensor, social, etc.) and tremendous volumes (yotta, yotta, yotta…) of business data at hand in the Big Data moment and realize insights

Second, following from this, Empirical Discovery is methodological a framework within and through which a great variety of analytical techniques at differing levels of maturity and from other disciplines are vetted for business analytical utility in iterative fashion by Data Science practitioners.

And third, it seems this vetting function is deliberately part of the makeup of empirical discovery, which I consider a very clever way to create a feedback loop that enhances Data Science practice by using Empirical Discovery as a discovery tool for refining its own methods.

Related posts:

Unsurprisingly, definitions of Big Data run the gamut from the turgid to the flip, making room to include the trite, the breathless, and the simply un-inspiring in the big circle around the campfire. Some of these definitions are useful in part, but none of them captures the essence of the matter. Most are mistakes in kind, trying to ground and capture Big Data as a ‘thing’ of some sort that is measurable in objective terms. Anytime you encounter a number, this is the school of thought.

Some approach Big Data as a state of being, most often a simple operational state of insufficiency of some kind; typically resources like analysts, compute power or storage for handling data effectively; occasionally something less quantifiable like clarity of purpose and criteria for management. Anytime you encounter phrasing that relies on the reader to interpret and define the particulars of the insufficiency, this is the school of thought.

I see Big Data as a self-defined (perhaps diagnosed is more accurate) condition, but one that is based on idiosyncratic interpretation of current and possible future situations in which understanding of, planning for, and activity around data are central.

Here’s my working definition: Big Data is the condition in which very high actual or expected difficulty in working successfully with data combines with very high anticipated but unknown value and benefit, leading to the a-priori assumption that currently available information management and analytical capabilties are broadly insufficient, making new and previously unknown capabilities seemingly necessary.

Related posts:

I’m late to making it available here, but O’Reilly media published the video recording of my presentation on The Language of Discovery: A Toolkit For Designing Big Data Interactions from last year’s (2012) Strata conference in NY.

Looking back at this, I’m happy to say that while my thinking on several of the key ideas has advanced quite a bit in the past 12 months (see our more recent materials), the core ideas and concepts remain vital.

Those are, briefly:

• Big Data is useless unless people can engage with it effectively
• Discovery is a critical and inadequately acknowledged aspect of sense making that is core to realizing value from Big Data
• Discovery is literally the most important human/machine interaction in the emerging Age of Insight
• Providing discovery capability requires understanding people’s needs and goals
• The Language of Discovery is an effective tool for understanding discovery needs and activities, and designing solutions
• There are known patterns and structure in discovery activities that you can use to create discovery solutions

I’ve posted it to vimeo for easier viewing – slides are here /user-experience-ux/strata-new-york-slides-new-discovery-patterns for those who wish to follow along – enjoy!

If you need such a deeper understanding of data science than Drew Conway’s popular venn diagram model, or Josh Wills’ tongue in cheek characterization, “Data Scientist (n.): Person who is better at statistics than any software engineer and better at software engineering than any statistician.” two relatively recent studies are worth reading.

‘Analyzing the Analyzers,’ an O’Reilly e-book by Harlan Harris, Sean Patrick Murphy, and Marck Vaisman, suggests four distinct types of data scientists — effectively personas, in a design sense — based on analysis of self-identified skills among practitioners.  The scenario format dramatizes the different personas, making what could be a dry statistical readout of survey data more engaging.  The survey-only nature of the data,  the restriction of scope to just skills, and the suggested models of skill-profiles makes this feel like the sort of exercise that data scientists undertake as an every day task; collecting data, analyzing it using a mix of statistical techniques, and sharing the model that emerges from the data mining exercise.  That’s not an indictment, simply an observation about the consistent feel of the effort as a product of data scientists, about data science.

And the paper ‘Enterprise Data Analysis and Visualization: An Interview Study‘ by researchers Sean Kandel, Andreas Paepcke, Joseph Hellerstein, and Jeffery Heer considers data science within the larger context of industrial data analysis, examining analytical workflows, skills, and the challenges common to enterprise analysis efforts, and identifying three archetypes of data scientist.  As an interview-based study, the data the researchers collected is richer, and there’s correspondingly greater depth in the synthesis.  The scope of the study included a broader set of roles than data scientist (enterprise analysts) and involved questions of workflow and organizational context for analytical efforts in general.  I’d suggest this is useful as a primer on analytical work and workers in enterprise settings for those who need a baseline understanding; it also offers some genuinely interesting nuggets for those already familiar with discovery work.

We’ve undertaken a considerable amount of research into discovery, analytical work/ers, and data science over the past three years — part of our programmatic approach to laying a foundation for product strategy and highlighting innovation opportunities — and both studies complement and confirm much of the direct research into data science that we conducted. There were a few important differences in our findings, which I’ll share and discuss in upcoming posts.

Related posts:

Discovery tools have had a referenceable working definition since at least 2001, when Ben Shneiderman published ‘Inventing Discovery Tools: Combining Information Visualization with Data Mining‘.  Dr. Shneiderman suggested the combination of the two distinct fields of data mining and information visualization could manifest as new category of tools for discovery, an understanding that remains essentially unaltered over ten years later.  An industry analyst report titled Visual Discovery Tools: Market Segmentation and Product Positioning from March of this year, for example, reads, “Visual discovery tools are designed for visual data exploration, analysis and lightweight data mining.”

Tools should follow from the activities people undertake (a foundational tenet of activity centered design), however, and Dr. Shneiderman does not in fact describe or define discovery activity or capability. As I read it, discovery is assumed to be the implied sum of the separate fields of visualization and data mining as they were then understood.  As a working definition that catalyzes a field of product prototyping, it’s adequate in the short term.  In the long term, it makes the boundaries of discovery both derived and temporary, and leaves a substantial gap in the landscape of core concepts around discovery, making consensus on the nature of most aspects of discovery difficult or impossible to reach.  I think this definitional gap is a major reason that discovery is still an ambiguous product landscape.

To help close that gap, I’m suggesting a few definitions of four core aspects of discovery.  These come out of our sustained research into discovery needs and practices, and have the goal of clarifying the relationship between discvoery and other analytical categories.  They are suggested, but should be internally coherent and consistent.

Discovery activity is: “Purposeful sense making activity that intends to arrive at new insights and understanding through exploration and analysis (and for these we have specific defintions as well) of all types and sources of data.”

Discovery capability is: “The ability of people and organizations to purposefully realize valuable insights that address the full spectrum of business questions and problems by engaging effectively with all types and sources of data.”

Discovery tools: “Enhance individual and organizational ability to realize novel insights by augmenting and accelerating human sense making to allow engagement with all types of data at all useful scales.”

Discovery environments: “Enable organizations to undertake effective discovery efforts for all business purposes and perspectives, in an empirical and cooperative fashion.”

Note: applicability to a world of Big data is assumed – thus the refs to all scales / types / sources – rather than stated explicitly.  I like that Big Data doesn’t have to be written into this core set of definitions, b/c I think it’s a transitional label – the new version of Web 2.0 – and goes away over time.

References and Resources:

• Inventing Discovery Tools
• Visual Discovery Tools: Market Segmentation and Product Positioning
• Logic versus usage: the case for activity-centered design
• A Taxonomy of Enterprise Search and Discovery