A New Approach to the Dynamic Organization of Knowledge

TERENCE C. HALLIDAY

The Internet poses a massive problem of making sense out of information. Two movements are taking place, side by side. On the one side, since the entry costs are very low, vast amounts of information of enormous variability in quality and value are being dumped on the virtual market. On the other side, to make some sense of it, "search entrepreneurs"—Excite, Yahoo, AltaVista—strive valiantly to create automated engines that will retrieve what we want when we want it. They succeed very imperfectly. One or two keywords are a very crude way to capture meaning of any complexity or nuance.	1
A Radical Solution
The National Institute for Social Science Information (NISSI), a nonprofit organization, has been founded by a group of scholars and civic leaders to pursue a radically different vision. Rather than create larger and larger data bases that are more and more difficult to search, NISSI proposes a different course: (1) select only premium knowledge, that is, the top 5 to 10 percent on a topic; (2) distill content to its essence for rapid comprehension; (3) rewrite technical material in regular English; and (4) integrate it into a meaningful knowledge "map."	2
This approach addresses several problems simultaneously. First, it does not assume that all knowledge is of equal value. Second, it does not assume that knowledge for experts is qualitatively distinct from knowledge for ordinary citizens. Third, it does not ask search engines to do what they are manifestly incapable of doing. Fourth, it respects the needs of most of us for efficiency and timeliness.	3
NISSI's approach creates and manages enriched, premium content, uniquely designed for the web, delivered on a breakthrough "Suggestion Engine" that simulates conversation. Take, for example, NISSI work on the topic of poverty. In partnership with Harvard University, NISSI has created an online library of the best research on poverty over the past ten years. First, NISSI formed a national panel of leading scholars on poverty, which selected 120 articles and books on four subtopics. Then NISSI's Information Analysts distilled these into several hundred "KeyTexts"—one- to two-page distillations of the main findings and arguments. These are loaded into a new kind of delivery platform—QuestWareTM. (See http://www.societyonline.org/urbanpoverty .)	4
QuestWare
A theory of cognitive science and artificial intelligence posits that individuals learn best when they are actively engaged in determining what they want to know when they want to know it. In casual conversation, each of us engages the other in a sophisticated exchange of questions and answers. Very quickly we zoom into the topics that interest us, and with the sophistication of a lifetime's experience, we efficiently find what it is the other person knows that we want to learn. Moreover, our questions have a kind of underlying logic we scarcely recognize.	5
QuestWare is a delivery platform that simulates conversation. It proceeds on the reasonable assumption that every oral comment or written text answers a set of questions just as it raises many others. QuestWare uses a delivery platform where Information Analysts have already anticipated the main questions answered by every text and a diverse array of questions raised by that text. We call this Question-Based Navigation. Every "KeyText" (a distilled summary of an underlying full text) has been content analyzed so that it is tagged with four to five Questions Answered and twenty to forty Questions Raised. An underlying technology matches the Questions Raised by one text with answers to those questions (Questions Answered) in other texts. As a result, the data base structure of content approximates a network, or as we call it, a "Knowledge Web." Of course, the same logic of Question-Based Navigation can also be applied to full-text and any other kind of content. All content answers questions and raises others.	6
When users navigate, it is as if they are having a conversation with texts. Users have a question. Answers to that question have already been anticipated by NISSI's Information Analysts. Users choose the answer, already provided, that most interests them. But that of course raises many other questions. Thus, users navigate not by use of keywords, but through suggested questions and answers that anticipate what they want to know. Of course, not all questions can ever be anticipated, so users can also use a conventional search approach for keywords. But that is a second-order mode of navigation, to be employed when the Suggestion Engine does not produce the results the user wanted.	7
The Suggestion Engine produced by QuestWare has many advantages.	8
When users come to a data base, they do not always know exactly what it is they want to know. QuestWare suggests many questions, including some that will be more interesting than the user originally had for him- or herself. Questions are a much more precise way of capturing underlying meaning than any concatenation of keywords. Questions are a more engaging way of gaining access to content and we are all extremely well versed in their use. By giving users twenty to thirty Questions Raised from every text, users have enormous flexibility and discretion in their mode of navigation. The system is very responsive to user interest.
Knowledge Webs, Questions, and Thinking Categories
All content can be given a structure or coherence. In consultation with expert panels, NISSI first creates a Content or Knowledge Map for each topical area. This gives users an overview of a field, a sort of table of contents of what they will find. Second, NISSI's Smart Libraries are organized in eight categories of questions that effectively offer an analytic schema for systematic inquiry in any context. Thus, for every text there are questions dealing with:	9
Context (what is the broader context in time and place? What are historical trends?) Details (what are examples, evidence, or definitions?) Causes (what brought about the situation in the KeyText?) Results (what are the consequences of the situation in the KeyText?) Alternatives (what are other theories and evidence?) Comparisons (what are differences in other times and places?) Outcomes (what are negative outcomes the author foresees?) Possibilities (what are policy or pragmatic recommendations or implications?)
By training ourselves or our students in these eight categories, we can develop a powerful heuristic for the critical approach to any body of content, whether it be scholarly articles or books, movies, newspaper articles, student papers, and so forth.
QuestWare and Legal History
What salience might these new ways of organizing knowledge have for legal history? What value does QuestWare provide for history's endeavor to create meaningful coherence from a body of underlying empirical material that relates to a problem of some sort over a period of time past?	10
Since I am a sociologist only lightly trained in history I shall speculate shamelessly.	11
1. Publishing
For those of us who write articles, books, or textbooks, the QuestWare logic creates a new opportunity for publishing.	12
In its print form, a textbook or book has a linear, fixed form. It is only possible to break out of this by either using a good index, which permits alternative reading "paths," or by skipping through the book, looking for chapter heads and subheads that seem to follow some theme of particular interest to the reader. In QuestWare, by contrast, the entire book can be turned into a dynamic knowledge web. The book is dismantled into segments. The segments are tagged with Questions Raised and Questions Answered. The book is then reconstituted as a knowledge web, where users may follow whatever thematic threads that most interest them, using the Question and Answer format. For those seeking examples, NISSI, in collaboration with Charles Sturt University in Australia, and Harcourt Brace (Australia), has reconstituted a text on evaluation in education using this method. Students spend more than 400 percent more time at this site, which also includes a threaded discussion forum, than they do at conventional educational web sites. Moreover, preliminary research indicates that the grades of distance learning students using this mode of education are significantly higher than regular classroom students.	13
Using the same network logic, a textbook can be surrounded by archival material and secondary sources, but systematically integrated with the original text using a question-based format. The segments of text are the core hubs in the system; they are linked by question categories both to other parts of the text and to original documents, and so forth, in a knowledge web. For example, the Harvard data base on poverty essentially arrays a vast amount of research around four chapters from William Julius Wilson's book, When Work Disappears (Knopf, 1997). KeyTexts from chapters form the "hubs" in the subtopical areas of Urban Community, Work, Family, and Economy. This is mostly invisible to the user, but it could be made much more visible—for research and teaching.	14
2. Teaching
In addition to the publishing approach we have just mentioned, it is also possible to organize course materials in knowledge webs on QuestWare. Imagine a course on the History of the Legal Profession in the U.S. A large number of primary and secondary sources (either full text or in KeyText form) can be integrated around topics, such as Recruitment to the Profession, Legal Education, the Social Organization of Professional Work, the Rise of Bar Associations, the Politics of Lawyers, the Demographic Transitions in the Profession. Of course, it is easy to link such materials using URL links to a multitude of sources inside or outside one's own site. To be sure, the QuestWare model, while labor intensive in the front end (that is, coding and embedding the questions and loading them into QuestWare), yields significant dividends for students (and for the R.A.s who actually do the work).	15
The teaching dividend is twofold: (1) Students are introduced to content that is richly integrated within itself and thus is more intrinsically satisfying to study and learn; (2) students are taught how to approach a body of archival material with a powerful analytical frame they can take with them for the rest of their lives. The potential is illustrated in a NISSI educational product called SocietyOnline® College Editions. This links course materials, lectures, textbooks, and a NISSI Smart Library via four kinds of educational activities (Smart Guides, Questercizes, Critical Thinking Tutors, Open Investigation). Explore at www.socieyonline.org/visitor/bc102 .	16
3. Archiving of Materials
Even more speculatively, when historians have responsibility for creating or reconfiguring an archive of electronic material, it may be possible to organize this material in QuestWare format. That is, full-text documents, or even entire files of documents (letters, transcripts, reports, and so forth) may be organized with QuestWare's two interlocking logics:	17
a. as a Knowledge Map, which is akin to normal methods of organizing information under a table of contents;	18
b. as a Knowledge Web, using Question-Based Navigation. Here documents and files would be linked by the Questions they answer and the Questions they raise. This would require more labor-intensive effort up front, but could yield great payoff downstream. Unlikely juxtapositions, threads of potential questions and answers, could lead to fruitful new questions and research agendas.	19
Imagine archiving materials on the (second) founding wave of bar associations in the U.S. between 1870 and 1920. The archives would include bar association documents and reports, legal newspaper and journals, personal correspondence of leading bar figures, as well as secondary literatures. For the Chicago Bar Association annual reports, for instance, it would be possible to ask Context questions (what trends in professional organization preceded the founding of the CBA in 1974?), Details questions (what kind of bar association was it? Elite? Populist? Voluntary? Compulsory?), Cause questions (what economic, political, cultural changes influenced the founding at this time?), Consequence questions (what impact did the founding have on its members, on other lawyers, on other associations?), Alternative questions (what are alternative theories of bar association founding?), Comparison questions (how was it different in other cities, counties, and states, in other periods?), and so forth.	20
Each of these questions would lead to other parts of the archive where answers could reasonably be expected to be found. If an archivist integrated these questions into a QuestWare format, it might rapidly advance intellectual debate and inquiry.	21
Of course, there is an implicit epistemology in these categories that requires careful judgment. Nonetheless, this approach could turn archives from repositories of essentially passive, classified information into dynamic systems of integrated primary sources.	22
Research and Research Centers
Does QuestWare and question-based organization of content have any relevance for the research enterprise? Perhaps. Here are five possibilities	23
a. In a literature review that often precedes a new research project, researchers and their research assistants could prepare KeyText versions of related texts—effectively as briefing materials—that could then form the hub of a Question-Based Navigation archive.	24
b. In the scrutiny of primary documents, researchers could collect copies of the most salient 20 percent and have them available for eventual integration into a "Smart Archive" that would be linked to their publications and (a) above.	25
c. For younger scholars especially, a systematic training in QuestWare's eight classes of questions would provide a simple analytic heuristic for their own research.	26
d. In publication of the research for the scholarly community, scholars could integrate (a) and (b) above with their publications. This would deepen the empirical support available for publications and provide a stronger evidentiary base for debate over interpretation.	27
e. In the featuring of research centers, many universities and history departments have centers that seek to define an intellectual niche in the scholarly community. Many centers are using the web for this purpose. The organization of content relevant to their web site's interests on QuestWare makes it significantly more valuable for many classes of users—other scholars, students, nonspecialists, and so forth.	28
Consider the following final example: The American Bar Foundation is a leading center for sociolegal research. However, much of this research is very technical and does not easily reach either nonspecialists or practitioners. The ABF contracted with NISSI to create a "Smart Library" on Law and Economic Discrimination. NISSI gathered some twenty articles and books produced by the ABF's economists of law—articles widely scattered through several disciplinary literatures, transformed them into KeyTexts, integrated them in a Knowledge Web on QuestWare, and linked them to the ABF's web site. Now key ABF constituencies—researchers, Fellows who provide financial support, lawyers, noneconomists, and nonspecialists—can easily learn answers to hundreds of questions that have been provided by ABF researchers. See at www.abf-sociolegal.org .	29
For an exchange of ideas, or further information about NISSI and its activities, contact Scott Parrott at sparrott@nissi.org . Or visit the NISSI web site at www.nissi.org .	30
Terence C. Halliday is Senior Research Fellow, American Bar Foundation, and President, National Institute for Social Science Information.

Content in the History Cooperative database is intended for personal, noncommercial use only. You may not reproduce, publish, distribute, transmit, participate in the transfer or sale of, modify, create derivative works from, display, or in any way exploit the History Cooperative database in whole or in part without the written permission of the copyright holder.