ITG Logo









Internetworking 3.3 Header

contents prev: Article-User Expectations next: Dear Aeby
ARTICLE

Designing for Information Foragers: A Behavioral Model for Information Seeking on the World Wide Web
James Kalbach, James.Kalbach@razorfish.de
Razorfish

Forward-thinking technology experts have predicted the day when accessing information, communicating with others, and performing complex transactions will be performed as thoughtlessly as turning on a light switch (Gelernter, 2000; Hawley, 2000). Computerized systems will satisfy our every information need. Information will ooze out of every corner of our lives.

This is, however, far from the current reality.

There are severe bottlenecks in the flow of information. Chief among these is poor interface design, which we have learned to accept with a shrug of the shoulders rather than demanding something better. When a system error occurs users even blame themselves: "I must have done something wrong," they say. This is particularly evident on the World Wide Web, where even basic interactions are unnecessarily complicated.

This paper explains and elaborates a behavioral model for understanding how people look for information on the Web. The first half briefly reviews a wide range key research to provide a broader context for understanding human information seeking behavior and a starting point for further exploration. The second part proposes a model for organizing design ideas based on this research.

The focus of this essay is limited to the World Wide Web for several reasons. The prolific growth of the Web, its yet untapped possibilities, and a general burgeoning excitement have secured its central role in information and communication services for the foreseeable future. The scope of this paper is focused on finding information on the Web and the behaviors people exhibit while performing this task.

Part 1

Information Seeking Behavior Research: A Background

Traditional paradigms of information retrieval tend to over-simplify the information seeking process. Robertson (1977) presents models in which the seeker simply enters a query and is given matching results (fig.1). Although this accurately describes part of the process, such models fail to take into consideration actual user needs, behaviors, situations and gaps in knowledge.



Fig. 1

After a wealth of post-war studies on human information behaviors, a clear paradigm shift in information research was declared (Dervin & Nilan, 1986). An explicit call for alternative, more holistic approaches to assessing information needs and uses was sounded. This has had a positive impact on subsequent research and opens fascinating possibilities for interface design.

New perspectives in research reversed traditional models of system design and placed the user at the center of attention, not the system. Three different, yet complementary research paradigms in particular mark this shift:

  • Since 1972 Brenda Dervin (1992) has developed a needs-based model she calls sense-making. Her methodology focuses on behaviors that people exhibit in their everyday experiences by analyzing a broad spectrum of complex human activity. The model aims at understanding user situations, gaps in knowledge, and the use of information.
  • Nicholas Belkin (1980) focuses on the notion that seekers, sometimes even experts in a given information system, are not able to properly formulate queries to access the information they need. He calls this "anomalous states of knowledge," or ASK for short. In his view, the failures in information retrieval lie on the system, not the seeker.
  • Robert Taylor (1984, 1985) developed a value-added approach for information seeking. He places the user’s problems at the center of attention and is concerned with linking problems with information traits of users. This model focuses on the perceived utility and value a user gets from a system and how this effects their decisions.

Seeking Information Online

Human information seeking behaviors in online settings present some unique problems and situations. Marcia Bates (1989) likens online seeking to berry picking: Seekers zigzag through a given information space, moving from resource to resource, varying search strategies rapidly. Within Bates’ berry picking model, browsing and searching are not seen as mutually exclusive activities, rather as complementary.

An important conclusion drawn from her research is the concept of the evolving search. As people look for information, the information need itself may change based on what is learned during the seeking process. Information seeking in online environments can therefore be seen as a negotiation between the searcher and the system. (Fig. 2)

Flexibility and user control while interacting with an information access system is therefore critical. Users must be able to interact naturally and intuitively with a system in order to meet their individual needs (Bates, 1990; Belkin, 1993).

Information seeking online: a negotiation between user and system

Similar to the berry picking model, Peter Pirolli and Stuart Card (1995) propose a theory of information foraging as an approach to analyzing human activities involving information access technologies. This theory is directly based on foraging theories in biology and anthropology in which humans are seen as adapters. The theory analyzes the trade-offs in the value of information gained against the cost of performing a task necessary to find information.

It is important to note that foraging for information does not equate to aimless "surfing." Foraging refers to the variety of strategies seekers exhibit in their quest for information and how humans adapt to their environments on a situational basis. Consequently, in an information-rich world the real design challenges are not only how to facilitate finding and collecting information, but how to optimize the seeker’s time.

Information Seeking on the World Wide Web

Jared Spool (Koman, 1998) has applied information foraging theory to web design with what he calls the "scent of information." In order to efficiently and effectively forage for information on the web, seekers need to have a sense of where they are going and why. The design of a navigation system should provide users with an accurate "scent" that they can follow to their destination.

Catledge and Pitkow (1995) were the first to publish a scientific study of Web browsing behavior. A chief finding of this research is that Web users rely on a limited number of pages within a site. The study clearly shows a recurring pattern of frequently returning to given page as a sort of "home base." The researchers describe this as a "hub and spoke" style of navigation through a web space.

Behavioral Models of Information Seeking

A seminal work in information behavior is Ellis’ (1989) development of a behavioral model of information seeking. This oft-cited framework has had a profound impact on information seeking research because it demonstrates patterns across situations and contexts. Six primary behaviors in information finding are identified:

  • Starting: Identifying relevant sources of interest
  • Chaining: Following and connecting new leads found in an initial source
  • Browsing: Scanning contents of identified sources for subject affinity
  • Differentiating: Filtering and assessing sources for usefulness
  • Monitoring: Keeping abreast of developments in a given subject area
  • Extracting: Systematically working through a given source for material of interest.

It is important to note that this categorization of behaviors does not indicate a uni-directional process for information seeking. Rather, the importance and involvement of each behavior in a given search is variable and situational. Actually occurrences of these behaviors can be, and usually are, iterative as well.

Marchionini (1995) proposes a similar model of the information-seeking process, tuned perhaps better to electronic environments. In his model, the information seeking process is composed of eight subprocesses which develop in parallel:

  • Recognize and accept an information problem
  • Define and understand the problem
  • Choose a search system
  • Formulate a query
  • Execute search
  • Examine results
  • Extract information
  • Reflect/iterate/stop (Marchionini, 1995; pp. 49-60).

Unlike Ellis (1989) this model rather describes a process than independent behaviors.

Researchers at the Carnegie Mellon University studied and organized user tasking on the Web, deriving what they refer to as a "Taskonomy," or logical groupings of task types (Byrne, et. al., 1999). They identify six main categories of user tasking:

  • Use information: Read, view, listen, save to disk, duplicate, print, etc.
  • Locate on page: Something "interesting," related concept, tagged information, etc.
  • Go to page: Hyperlink, back, forward, bookmark, history list, provide URL, etc.
  • Provide information: Search string, shipping address, survey response, etc
  • Configure browser: Add bookmark, set helpers, change cache size, scroll, etc.
  • React to environment: respond to dialog, respond to display change, reload, etc.

These in part resemble Ellis’ and Marchionini’s information seeking behaviors, however, add some tasks that are specific to the web.

More recently, Chun Wei Choo and Don Turnbull (2000) examined information seeking on the web. They developed a integrate model of browsing and searching based on established research mentioned here, specifically Ellis’ (1989). Wilson’s (1997) modes of browsing and searching are also included in the model: undirected viewing, conditioned viewing, information searching and formal searching. Choo and Turnbull combine these modes with Ellis’ behavioral model to arrive at the following matrix for characterizing behavior:

 

Behavioral Modes and Moves of Information Seeking on the Web

 

Starting

Chaining

Browsing

Differentiating

Monitoring

Extracting

Undirected Viewing

Identifying, selecting, starting pages and sites

Following links on initial pages

       

Conditioned Viewing

   

Browsing entry pages, headings, site maps

Bookmarking, printing, copying;
Going directly to known site

Revisiting 'favorite' or bookmarked sites for new information

 

Informal Search

     

Bookmarking, printing, copying;
Going directly to known site

Revisiting 'favorite' or bookmarked sites for new information

Using (local) search engines to extract information

Formal Search

       

Revisiting 'favorite' or bookmarked sites for new information

Using search engines to extract information

(Texts are meant to provide short, generic examples of that behavior.)

While this research is not without its flaws, the researchers successfully map Ellis general behavioral framework onto information on the Web.

Summary

Main assumptions drawn from the all of the above research are:

  • Failures in information retrieval are caused by the result of poor system design, not user error or inabilities (Dervin 1986, Belkin 1980, Taylor, 1983, 1984).
  • Online search strategies change rapidly and searches evolve as users learn more about what is available within a system (Bates, 1989).
  • User must be able to properly interact and negotiate with systems in order to meet their information needs (Bates 1990, Belkin, 1990).
  • In information rich environments searchers constantly weigh the potential information gained against the cost of performing a task necessary to find information. Therefore, an important design challenge is optimizing users time (Pirolli, 1995).
  • Information seeking on the web has some unique attributes, but can for the most part be explained by traditional models of behavior (Choo, Turnbull, 2000).

Part II

Designing for information foragers

In most disciplines and fields of study there is a significant gap between what is known theoretically and what is applied practically. One problem with some research is that results often remain relatively unknown and do not inform practice. Much needed is a tactical exploitation of research findings to arrive at successful design solutions.

The rapid development of the Web has wrought on significant changes in how design consultancies think about and carry out design. Interaction designers face constant pressure not only to adapt to changing technologies, but also to accommodate the needs and expectations of demanding and impatient users. For Web developers and designers, there is an increased need for design strategies and techniques that are flexible and efficient to employ.

Below is a summary of some of the information seeking behaviors developed in the above-mentioned research with recommendations for facilitating the corresponding behavior into a design. The ultimate goal is to provide users with information that could potentially help them determine the relevance of a website in relation to their specific information need.

The examples below demonstrate the rationale for creating information finding mechanisms based on general seeking behaviors on the Web. It assumes a generic, "all purpose" website. In reality some websites might have a stronger demand of supporting one behavior over another. In some situations one aspect may not be important at all.

A general information portal on the Web, for example, must clearly support the starting, linking and browsing, but not necessarily monitoring or extracting. A car-buying site, on the other hand, might need to allow users to extract detailed information and monitor information over months during the purchase decision-making process. Use of the proposed framework should therefore be scaled to a given situation.

The behaviors used here stem from Ellis’ (1989) original model, which Choo and Turnbull (2000) have tested and supported. For the sake of using standard web technology, "chaining" is called "linking" here. In addition, browsing is broken into "browsing" and "searching", which according to Bates (1989), are part of a related strategic move. Finally, this model is not comprehensive, but rather a starting point intended to stimulate thinking towards design solutions that facilitate user information behaviors. The intent is not to limit or dictate design or design standard, but to serve as the bases for making decisions.

1.0 Behavior : Starting

This refers to identifying relevant sources on the Web. There are four principle ways in which users arrive at given website:

    • Typing in a URL directly
    • Referring to a bookmarked URL
    • Following a link from another site
    • Using a search engine

1.1. Example features to support Starting:

    • Thoughtfully-designed URLs that are easy to remember
    • Alternate spellings for URLs to anticipate typos
    • Unique URLs for each page
    • If URLs are generated from a database, they should be converted to static URLs
    • Page construction that allows for easy and accurate bookmarking.
    • Carefully worded page titles that provide a useful context.
    • Appropriate use of meta tags to describe the site. This not only facilitates indexing by search engines, but also yields meaning descriptions in search engine results lists.

In considering Starting as a behavior, it is important to remember that users probably first come in contact with a given website from a referring source, e.g. another website or elsewhere, such as advertising and other referring sources. The decision to go to visit one website over another is often based on micro-content found in the URL, page title and meta-tags.

2.0 Behavior : Linking

Linking is the act of following and connecting new leads found on initial websites. Chances are good that when users arrive at a given site, they are not on the page they need to be. They must be able to orient themselves quickly, often within seconds, and determine which links to follow.

2.1 Example features to support Linking:

  • Accurate, descriptive and mutually exclusive link names that make sense to visitors in terms that they can relate to and understand
  • Consistent navigation through design elements such as placement, style and general look and feel
  • Sense of orientation created with page titles, global elements, consistent use of color and graphics
  • A way out or back - Generally this means not disabling the back button or using unnecessary extra windows or new browsers
  • Clearly marked links that
  • Icons with clear meanings
  • A limited number of well-organized navigation options – 7 options is generally accepted as standard, although research shows that fewer that 5 is even better.
  • Navigation that provides the appropriate, relevant associations related to page content to anticipate users’ probable next moves.

3.0 Behavior : Browsing

Browsing means scanning site contents and informally grouping items by subject affinity. This is a behavior that is hard to image NOT occurring in every Web session. Browsing is a chief Web activity. Supporting this behavior in web design is essential.

3.1 Example features to support Browsing:

  • Meaningful categories that themselves convey a message and the purpose of the site
  • Prioritized navigation – separating navigation into meaningful types to facilitate browsing of options.
  • Clearly presented and readable text
  • Content overview – the Web offers the possibility to "chunk" content. That is, it is not necessary to present all content at once, rather in digestible pieces that provide a clear overview.
  • Bulleted lists, tables and other constructions that facilitate scanning

4.0 Behavior : Searching

This refers to direct and targeted searches using a search engine or similar functionality. Important considerations here are twofold: the design of the search interface and functionality, and display of search results.

4.1 Example features to support Searching:

    • Allows users to limit in meaningful ways – standard operators (AND, OR and NOT) should be available, in addition to others
    • Query syntax is standard or easily learned - HotBot.com offers limiters nicely with pull-down menus
    • Search results provide a context for understanding hits, such as page title, date, description – Google.com even shows part of the sentence in which the search string was found
    • Opportunity to revise search and to search again
    • If no hits, suggestions to similar, possible sources are presented
    • Spell check with corresponding suggestions for "correct" spellings

5.0 Behavior : Differentiating

Differentiating is the act of evaluating information for relevance to the information need or problem. In some was this is the combined goal of the above-mentioned behaviors and features. Additional features, however, can directly help visitors uncover the value a website has to offer. Designers should strive to strike a balance between control and freedom.

5.1 Example features to support Differentiating:

    • Logical and meaningful headings that explain content to some degree
    • Appropriate text lengths that are suitable for reading online
    • Chunking content into layers and allowing random access into different sections of information
    • Summary texts and abstracts that indicate the quality, usefulness and scope of content
    • Providing deeper content for those who need it
    • Contact information and help to assist people who need more information or who couldn’t locate exactly what they need
    • User comments and reviews of websites, Web content or products sold over the Web
    • Indications of and links to semantically related material

6.0 Behavior : Monitoring

Monitoring is often not a user need and not a site goal. This means keeping abreast of developments in a given field. Deeper and ongoing information needs often cannot be fulfilled in one online session. Some sites, however, overlook this behavior and could benefit by supporting it.

Example features to support Monitoring:

  • Easily bookmarked pages for quick and easy access at a later date
  • Newsletter subscription
  • Email alerts that notify users of changes in content, current status, updates, etc.
  • SMS message to communicate up-to-the minute changes and breaking news
  • Online agents that collect, control and communicate information and changes
  • Customized pages that allow users to configure site elements to their liking
  • Personalized page that react dynamically to user activity

7.0 Behavior : Extracting

As the name implies, extracting refers to taking and using the appropriate, identified information or pieces of information online. It is the final use of information.

7.1 Example features to support Differentiating:

  • The ability to print; print friendly formats, if needed. Dark colored page backgrounds and frames complicate printing greatly
  • Cut and paste as an option – this means providing key bodies of information in HTML format. Other formats, such as images or flash, do not allow for cut and paste.
  • Download possibilities – Portable document files (PDF) have become standard and readers for this format are ubiquitous and free. Other formats for download can also be considered depending on use and target groups
  • Applications as filters of large bodies of information
  • Sorting functions

 

Although the framework presented above is focused on and limited to websites, it can be transferred on to other media formats. It is meant to provide the rationale for making design decisions.

Missing from this framework are two important issues that require separately handling:

  1. Identifying and explain the motivations and expectations that users have before coming to the Web to solve a problem.
  2. Investigating the overall information ecology or contexts in which users behave.

 

Conclusion

The Information Revolution has only just begun. Understanding human behaviors in accessing information has never been more crucial. Because of its wide-reaching impact and present central role, ideas currently being implemented on the Web will have lasting effect on future information access systems, platforms and environments.

Future design must strive to create intelligent and intuitive systems that engage the user in meaningful negotiations with information on their own terms. In the end the success or failure of a given solution lies not on new technologies, but on understanding human needs and behaviors. This paper is a call to develop more intelligent, intuitive systems that are user-centered and support fundamental human information needs and goals.

 

Bates, Marcia. J. (1989). The design of browsing and berrypicking techniques for the online search interface. Online Review, 13, 407-424.

Bates, Marcia. J. (1990). Where should the person end and the information search interface start? Information processing and management, 26(5): 575-590.

Belkin, Nicholas J. (1980). Anomalous states of knowledge as the basis for information retrieval. Canadian journal of information science, 5, 133-143.

Belkin, Nicholas J., Marchetti, P.G., & Cool, C. (1993). BRAQUE: Design of an interface to support user interaction in information retrieval. Information processes and management, 29(3): 324-344.

Byrne, Mike. D, John, B. E., Wehrle, Neil. S., & Crow, D. C. (1999). The tangled web we wove: A taskonomy of www use. Human factors in computing systems: Proceedings of CHI 99: 544-551. Reading, MA: Addison Wesley.

Choo, Chen Wei, & Turnbell, Don (2000). Information seeking on the web: An integrated model of browsing and searching. FirstMonday, 5(2). (http://firstmonday.org/issues/issue5_2/choo/index.html).

Ellis, David. (1989). A behavioural model for information retrieval system design. Journal of information science, 15 (4/5): 237-247.

Dervin, Brenda. and Nilan, M. (1986). Information needs and uses. Annual review of information science and technology, 21: 3-33.

Dervin, Brenda. (1992). From the mind's eye of the user: The sense-making qualitative-quantitative methodology. In J. D. Glazier & R. R. Powell (Eds.), Qualitative research in information management (pp. 61-84). Englewood, CO: Libraries Unlimited.

Glenter, David. (2000). The second coming – a manifesto. Edge (www.edge.org/documents/archive/edge70.html).

Hawley, Michael. (2000). Life after the internet. Speech given at Business WeekMagazine’s "Technology Summit," May, 2000. (summarized in Ochman, B.L. http://www.webeview.com/2000/05/19/feature/index01.html.)

Koman, Richard. (1998). Helping users find their way by making your site "smelly". WebReview.com (http://www.webreview.com/pub/98/05/15/feature/index.html).

Marchionini, Gary N. (1995). Information seeking in electronic environments. Cambridge, Eng.: Cambridge University Press.

Pirolli, Peter, & Card, Stuart. (1995). Information foraging in information access environments. Human factors in computing systems: Proceedings of CHI 95. (http://www.acm.org/turing/sigs/sigchi/chi95/Electronic/documnts/papers/ppp_bdy.htm)

Robertson, S. E. (1977) Theories and models in information retrieval. Journal of documentation, 33: 126-148.

Taylor, Robert. S. (1984). Value-added processes in document based systems: Abstracting and indexing services. Information services and use, 4(8): 127-146.

Taylor, Robert. S. (1985). Information values in decision contexts. Information management review, 1(1): 47-55.

Wilson, T. D. (1997). "Information behaviour: An interdisciplinary perspective," Information processing & management, 33 (4): 551-572.

Worksheet

Developing Information Seeking Support Mechanisms

 

1.0 Preliminary Considerations

1.1 Project goals:

1.1.1

1.1.2

1.1.3

1.1.4

1.1.5

1.1.6

1.1.7

 

 

1.2 Target group anaylsis - List target groups in order of priority for project.

Target Group

Description

Information needs

Information behaviors

1.2.1

 

 

 

 

   

1.2.2

 

 

 

 

   

1.2.3

 

 

 

 

   

1.2.4

 

 

 

 

 

 

1.2.5

 

 

 

 

 

 

1.2.6

 

 

 

 

 

 

1.2.7

 

 

 

 

 

 

 

1.3 Information types and formats that support the above needs and behaviors:

Usage: need or behavior

Supported by

Information type / format

1.3.1

 

   

1.3.2

 

   

1.3.3

 

   

1.3.4

 

   

1.3.5

 

   

1.3.6

 

   

1.3.7

 

   

 

 

2.0 Competitor / Related Site Audit

Competitor

Special feature / Innovative solutions

Behavior(s) supported

 

 

   

 

 

 

   

 

 

 

   

 

 

 

   

 

 

 

   

 

 

 

   

 

 

 

3.0 Information Seeking Mechanisms

Behavior

Priority for project

Supported by:

Relevant to user group:

Starting

     
       
       
       

Linking

     
       
       
       

Browsing

     
       
       
       

Searching

     
       
       
       

Differentiating

     
       
       
       

Monitoring

     
       
       
       

Extracting

     
       
       
       

 

  1. Evaluation

Estimations of success for any given information search should be made. This can be accomplished with a range of methods, including:

  • Review sessions
  • Role playing
  • User scenario walkthroughs
  • Naming conventions review and testing
  • Prototype testing
  • Usability testing
contents prev: Article-User Expectations next: Dear Aeby

© Internet Technical Group
Last update: January 27, 2001
URL: http://www.sandia.gov/itg/newsletter/dec00/article_information_foragers.html