Mary E. Brown, Ph.D., Professor
Information Science

Southern Connecticut State University
501 Crescent Street, New Haven, CT 06515

Department of Information and Library Science
Fax: 1.203.392-5780 / Phone: 1.203.392-5781
Toll Free: 1-888-500-SCSU, then press 4

In this unit we ask two questions: Can the user be the weak link in a Digital Library? How can you get a sense of a DL collection? In answering this question we consider human attention as a limited resource and phrase-based browsing and lexical navigation.

CAN THE USER BE THE WEAK LINK IN A DIGITAL LIBRARY?

Human attention: a limited resource

Levy (1997) considers the issues of reading and attention with applications to the digital environment. As Levy points out, there is a growing awareness that human attention is a highly limited resource: In the context of our information-enriched society, this presumably immutable limitation needs to be well understood. Levy recalls from the early psychologist William James (1890) that attention is "the taking possession of the mind, in clear and vivid form....Focalization, concentration, of consciousness are of its essence. It implies withdrawal from some things in order to deal effectively with others...." (p. 203 in Levy). In the words of noted cognitive psychologist Herb Simon, "In a world where attention is a major scarce resource, information may be an expensive luxury, for it may turn our attention from what is important to what is unimportant. We cannot afford to attend to information simply because it is there" (p. 203 in Levy).

Observation of attention have lead to what W. Thorngate calls the economics of attention. Three of Thorngate's principles of attention economics are: "attention is a finite and non-renewable resource" (principle of fixed attentional assets ); "attention can, in general, be invested in only one activity at a time" (principle of singular attentional investments); and "whenever we search for and choose attentional investments, the acts of searching and choosing themselves require attentional investments" (principle of exploratory attentional expenses ) (p. 203 in Levy). To summarize, we have a fixed amount of attention, it can be directed toward one activity at a time; the act of choose among activities, itself, uses up attention. The overall effect in an information busy environment: we may expend our attention on the single activity of trying to choose which other activity needs our attention. (The Catch 22 of the information society.)

It is of interest the relationship between the reading material and our attention. To look at this, we will review Levy's brief coverage of the history of reading in the dominant western world. Prior to the Middle Ages, reading was primarily a read-aloud activity (whether there was a listening audience or the reader was reading aloud to his or her self). During the Middle Ages, reading shifted to a primarily silent reading activity. Following the invention of the printing press, and the resultant explosion in the availability of books, there was an anxiety created by the sudden increased awareness of the amount of knowledge and the lack of an organization of its various fragments. Between the end of the Middle Ages and the eighteenth century, an immense effort was put into creating vast organizational schemes and institutions to oversee these schemes. The expanded availability of books and the shift to silent reading, saw the development of intensive reading, that is, a rehearsed and varied knowledge obtained from rereading of familiar material. The second half of the eighteen century, with expanded book production, lending libraries, and newspapers, saw a shift to extensive reading habits, that is, the critical and avid reading of great breadths of material. The design of books, including typefaces, spacing, margins, and binding, aided the extensive reading process. More specifically, the design of books guided and focused our attention during extensive reading.

We have inherited the tradition of extensive reading. However, current trends in production of reading materials--from limited resolution screens to beeping and humming computers--work against the attention that is necessary to sustain extensive reading. Visually-busy displays further compete for our attention. If digital libraries are to contribute to enhancing how we deal with information, designs for digital libraries must take into consideration what we know about how we read and to what we can attend.

Talking in libraries

Crabtree, Twidale, O'Brien and Nichols (1997) are concerned with the methodologies used in systems analysis and design, specifically in designing digital libraries. Crabtree et al. describe and justify their use of ethnomethodologically-informed ethnography to study query negotiation at the help desk of a university library. Crabtree et al. Advance that the success of systems design depends on the social context into which the system will be placed. That is, to achieve success, the system must fit and complement the social character of the work setting. Ethnomethodologically-informed ethnography involves gathering descriptions of what is being done and the ways in which it is being done. This information must be gathered in the "real world" setting and activity. Talking in libraries

In ethnomethodologically-informed ethnography, descriptions are gathered using a documentary method. That is, through recording the action on audio-tape (or video-tape, if appropriate) and analyzing the transcribed actions and interactions. This methodology permits the researcher to ask task-based questions of actors during the taping. Appropriate questions include What are you doing? and Why did you do that?

This methodology helps uncover cooperative and sometimes subtle aspects of work which often go undiscovered using other methodologies. While the points made in this article are well taken, they are not well stated. The following work, notwithstanding Crabtree et al.'s criticism, is recommended for additional insight into the issue raised by these researchers:

Taylor, R. S. Question-negotiation and information seeking in libraries. College & Research Libraries 29, 3 (1968) pp. 178-194.

Works by Peter Ingwersen on interpersonal communication and library science are also worth reading, especially:

User-librarian negotiations and information search procedures in public libraries: Analysis of verbal protocols. ED211051. (Soren Kaae, co-author).

Search procedures in the library--analyzed from the cognitive point of view. Journal of Documentation 38, 3 (1982) pp. 165-191.

HOW CAN YOU GET A SENSE OF A DL COLLECTION?

Phrase-based browsing

In a physical collection, such as a traditional library, the user can gain a sense of the coverage of the collection by browsing through its stacks. Browsing, in fact, has been shown through research to be "a rich and fundamental human information behavior." How does the user of a digital library gain this same sense of the collection? Nevill-Manning, Witten and Paynter (1997) respond to their own question with Sequitur, an algorithm for building hierarchical structures of frequently repeated phrases in a collection in order to browse the content of a digital collection.

Sequitur first displays to the user a list of words derived from the lexicon of the digital collection. The user then selects a word. The selected word is next shown imbedded in frequently occurring phrases derived from the collection. Finally, the selected phrase is shown imbedded in the longer phrases in which it occurs in the various items in the collection. Through this process, the user can gain a sense of the topics covered in a given collection. The organization of the phrases are similar to indexes based on permuted titles or keywords-in-context (KWIC). Examples of the resulting displays are:

words in the collections lexicon:

index

indexed

indexer

words in phrases:

representation indexing

text indexing

indexing and retrieval

phrases in longer phrases:

all indexing and retrieval of multimedia

text indexing and retrieval document processes

Nevill-Manning et al. point out that phrase browsing shifts attention from the individual item in the collection to the content of the collection as a whole and permits users to gain a good sense of the subject matter covered in a large collection while presenting the information in manageable chunks.

Problems identified by Nevill-Manning et al. in initial tests include artificially high frequencies for phrases quoted in the abstract from the text by the author and for title repeated at the top of each text page. The technique requires a large amount of memory and creates technical problems in collections larger than 50 Mbytes. (See Nevill-Manning et al. for details of the Sequitur algorithm and phrase identification.)

Lexical navigation

Cooper and Byrd (1997) developed an interface which would allow a user to expand or refine a query based on the actual content of a collection. Cooper and Byrd developed, for use in the interface, a technique they named lexical navigation . The LexNav document query and retrieval system addresses a problem cited in the literature: the problem of user-choice of vocabulary. In translating their concepts of their information needs into query terms that will match terms found in the collection being queried, users often generate terms that are 1) not at the proper level of specificity (for example, "fruit" instead of "grapes," "pears," or "oranges") or 2) are ambiguous (for example, "nut," which can be a food, a hardware item, or an eccentric).

LexNav extracts from a collection three things: keywords, a context thesaurus, and a lexical network of relationships. To conduct a search, the user enters a query term into a multi-line text box, then clicks one of two buttons: "view document titles" or "view related concepts." The choice to "view related concepts" calls the context thesaurus and related terms are displayed in a separate text window. The user may add a related term to the query or click to see the relationships between search terms (related terms called from the context thesaurus) and nearby terms (included other related terms). Initially, relationships are displayed in three columns labelled: "search term," "how related," and "is related to." The relationship between terms in columns one and three are shown either as a named relationship (for example, "location," "chairman and chief executive") or as a strength of the relation (represented by one, two, or three asterisks). The user may choose to see term relations displays as a graphic network. On the network graph, terms which can be expanded further (to show additional relationships) are highlighted.

Once the user is satisfied with the terms in the query window, titles of found documents can be displayed in a separate text window. Selecting a title opens a new page containing three windows: one window in which the query terms are highlighted in a KWIC (keyword in context) display of the document; one window in which keywords found in the document are listed; and one window in which the document itself is displayed. By clicking, keywords in the document are highlighted. By selecting any line in the KWIC display, the document display is brought to that line in the text.

Certainly, this is a creative interface which will permit the user to discover situated query terms to fit to their information need and will permit the user to move more effectively to promising portions of a document. However, the researchers claim that this interface is one for inexperienced users is questionable, unless inexperienced applies to context expertise in the given collection but does not apply to information technology skills. LexNav runs as applets embedded in Web pages.

REFERENCES

Cooper, J. W. & Byrd, R. J. (1997). Lexical navigation: Visually prompted query expansion and refinement. Proceedings of the 2nd ACM International Conference on Digital Libraries, 237-246.

Crabtree, A., Twidale, M. B., O'Brien, J., & Nichols, D. M. (1997). Talking in the library: Implications for design of digital libraries.Proceedings of the 2nd ACM International Conference on Digital Libraries, 221-228.

Levy, D. M. (1997). I read the news today, oh boy: Reading and attention in digital libraries. Proceedings of the 2nd ACM International Conference on Digital Libraries, 202-211.

Nevill-Manning, C. G., Witten, I. H., & Paynter, G. W. (1997). Browsing in digital libraries: A phrase-based approach. Proceedings of the 2nd ACM International Conference on Digital Libraries, 230-236.