IRList Digest Saturday, 7 January 1989 Volume 5 : Issue 6 Today's Topics: Abstracts - Dissertations selected by S. Humphrey News addresses are Internet: fox@vtopus.cs.vt.edu BITNET: foxea@vtcc1.bitnet (replaces foxea@vtvax3) ---------------------------------------------------------------------- Date: Thu, 8 Dec 88 20:57:22 EST From: "Susanne M. HUMPHREY" Subject: some IR-related abstracts AN University Microfilms Order Number ADG88-11084. AU LANGLE, GERNOT-BERNHARD. IN University of Minnesota Ph.D 1988, 242 pages. TI AN INVESTIGATION OF THE EFFECT OF SPECIFIC KNOWLEDGE IN FUNCTIONAL AREAS OF BUSINESS ON INFORMATION SYSTEMS ANALYSIS AND DESIGN. DE Business Administration, General. AB Recent studies on the design of computer based information systems have indicated that the content of the knowledge base and the reasoning behavior of systems analysts are two important factors in the development of computer based information systems. This study focuses on knowledge which relates to a specific functional area of business (such as accounting, manufacturing, or marketing) and the effect of the presence or absence of such function specific domain knowledge on how systems analysts determine information requirements. Determining information requirements is postulated to involve the construction of representations utilizing modeling, discovery, and validation processes. The specific study was an empirical investigation involving eight experienced business systems analysts. Each subject was asked to analyze and design two information systems, one in a familiar area of business (one about which the subject possessed function specific domain knowledge), the other in an unfamiliar area of business. Subjects were instructed to think out loud during the tasks and their verbalizations were tape recorded and transcribed for analysis. The results indicate that the presence of function specific domain knowledge affected the construction of representations as well as modeling, discovery, and validation processes. Subjects with function specific domain knowledge were found to (1) build representations considering a larger number of facts and concepts relating to the information system's application domain; (2) discover and validate the representations requesting additional domain specific information more frequently; and (3) model the information system utilizing analogical reasoning more often than subjects without function specific domain knowledge. Interpretation of the results suggests that when experienced systems analysts possess function specific domain knowledge, they are more likely to construct accurate and complete systems representations and to exhibit focused and efficient information gathering behavior. This results in increased efficiency in information requirements determination, or in increased quality of information requirements specifications, or both. AN University Microfilms Order Number ADG88-01493. AU MURALIKRISHNA, M. IN The University of Wisconsin - Madison Ph.D 1988, 170 pages. TI OPTIMIZATION OF MULTIPLE-DISJUNCT QUERIES IN A RELATIONAL DATABASE SYSTEM. DE Computer Science. AB In this thesis, we describe the optimization of arbitrarily complex queries expressed in relational calculus. The qualification list is allowed to be any complex boolean expression involving both ANDs and ORs. In other words, the qualification list may have an arbitrary number of disjuncts. The query graph of each disjunct may also have any number of components. Optimizing the various disjuncts independently of each other can be very inefficient. Considerable savings in cost can be achieved by optimizing the various disjuncts together. In a multiple-relation multiple-disjunct query, it may be possible to combine two or more disjuncts into one term. This will cut down the number of scans on each relation and also the number of times each join is performed. The objective will be to merge the disjuncts into the minimum number of terms. Minimizing the number of terms can be formulated as the problem of covering a merge graph with the minimum number of complete merge graphs, which are a restricted class of cartesian product graphs. The problem of minimizing the number of terms is NP-complete. We present polynomial time algorithms for special classes of merge graphs. We provide a heuristic for general merge graphs. For single-relation multiple-disjunct queries involving more than one attribute, an optimal access path might consist of more than one index. The cost in our optimization model, for single relation queries, is measured in terms of the number of pages fetched from disk. We will formulate the problem of finding a set of optimal access paths for a single-relation multiple-disjunct query as one of finding a minimum weighted vertex cover in a hypergraph. Finding the cheapest vertex cover in a hypergraph is NP-complete. We present a new approximation algorithm that gives near optimal vertex covers for random hypergraphs over a wide range of edge probabilities. We also demonstrate the usefulness of equi-depth multi-dimensional histograms in optimizing queries using multi-dimensional indices. AN University Microfilms Order Number ADG88-03029. AU EDYBURN, DAVE LEE. IN University of Illinois at Urbana-Champaign Ph.D 1987, 270 pages. TI AN EVALUATION OF THE INFORMATION RETRIEVAL SKILLS OF STUDENTS WITH AND WITHOUT LEARNING HANDICAPS USING PRINTED AND ELECTRONIC ENCYCLOPEDIAS. DE Education, Technology. AB The use of online databases for information retrieval has risen substantially in recent years as a result of issues associated with accessibility, convenience, and efficiency. This study examined student use of information technology and focused specifically on students' natural information retrieval skills by (a) comparing student use of printed and electronic forms of an encyclopedia, (b) examining the effects of menu or command driven interfaces and task difficulty on students' success of retrieval of specific factual information using encyclopedias, and (c) assessing students' attitudes toward information retrieval tasks when using printed and electronic encyclopedias. Thirteen junior high students with learning handicaps and fifteen students without learning handicaps were randomly selected and randomly assigned to one of three treatment orders in a repeated measures Latin square design. Assessments were made on each student's IQ, spelling ability, reference skills, and keyboarding speed. Each treatment session involved retrieving specific factual information on four search tasks using one of three encyclopedias: print, electronic with menus, or electronic with commands. Students completed an attitude scale at the end of every search session. Significant differences were found between the two groups' levels of success on all retrieval tasks. The menu driven electronic encyclopedia was found to improve retrieval success significantly over the command driven version. However, the menu electronic encyclopedia was not found to improve retrieval success significantly over the printed encyclopedia. Further, the command driven encyclopedia significantly impaired retrieval success to a level lower than that in the printed encyclopedia. Significantly greater retrieval success was also found on assigned tasks versus self-selected tasks and on simple tasks versus complex tasks. Regression analysis indicated that scores from the reference skills test and keyboarding speed were significant predictors of retrieval success, whereas IQ and spelling ability were not. Students' attitudes toward information retrieval tasks were not affected by the type of encyclopedia used, but students with learning handicaps were found to hold significantly higher (positive) attitudes than their nonhandicapped peers after all three treatments. The implications of the results for teaching online searching to students are noted. AN University Microfilms Order Number ADG88-07814. AU LINN, RICHARD JOHN. IN The Pennsylvania State University Ph.D 1987, 164 pages. TI AN EXPERT SYSTEM APPROACH TO THE CONTROL OF AN AUTOMATED STORAGE/RETRIEVAL SYSTEM. DE Engineering, Industrial. AB An architecture for an Expert System based control system for the control of an Automated Storage and Retrieval System (AS/RS) is created. The control system includes several unique elements, including the hierarchical partitioning of the control and decision tasks as well as the use of simulation to predict system performance. Strategic control policies and decisions are made utilizing a knowledge base of strategies that have been shown to perform well in practice. Tactical planning includes implementation of strategic plans during normal operation periods. The tactical planning module also includes a mechanism that detects input congestion and potential or existing system blockage. Alternative cart control scenarios are suggested by an expert system based tactical decision module. These control scenarios are then run through a short-run deterministic simulation in order to determine their merit. The control that best affects the system is implemented. The system was compared to the conventional system, and the experimental result has shown that the expert system based control appears to be an excellent control alternative for an AS/RS operation. AN University Microfilms Order Number ADG88-07364. AU THABIT, HASSAN J. IN University of Pittsburgh Ph.D 1987, 187 pages. TI INFORMATION TRANSFER: THE DIFFUSION OF MICROCOMPUTERS IN SAUDI ARABIAN UNIVERSITIES. DE Information Science. AB The study's purpose was to examine the diffusion of micros in Saudi Arabia in relation to hardware, software, users, and protential users. The study investigated the micro use by university students and faculty and identified available micro hardware and software. The study determined the source and channels of communications used and needed for the diffusions of micros and examined the vendor's role as information provider. Factors affecting the diffusion were investigated and the overall diffusion was evaluated as to: relative advantages, compatibility, complexity, trialability, and observability. Questionnaires and interviews were used, and the subjects were students and faculty members, and micro vendors. The findings showed that micros were perceived to have relative advantages, compatibility, simplicity, trialability, and observability by the study's social systems. Of the students and faculty surveyed, 28.6% used micros and the social network for micros was found to be generally established. However, vendor communication with clients was perceived as not very effective even though clients perceived vendors to have empathy and homophily. Four factors were perceived to hinder micro diffusion: limited availability of Arabic micro literature; limited availability of university courses about micros; limited services including maintenance, training, and consultation; and lack of public awareness. Conversely, the following problems were found: abnormal, vague appearance of Arabic characters on the screen; lack of Arabic software; confusing Arabic computer terminology; and difficulty programming in Arabic. The absence of copyright policy for software was also determined to affect the development of software. AN This item is not available from University Microfilms International ADG05-62889. AU HOPKINS, RICHARD LEONARD. IN University of Toronto (Canada) Ph.D 1988. TI THE INFORMATION-SEEKING BEHAVIOUR OF LITERARY SCHOLARS IN CANADIAN UNIVERSITIES. DE Library Science. AB The descriptive objective of the study was to test nine assumptions about the information seeking behaviour of Canadian literary scholars. The analytical objective was to explain different degrees of campus library use by these scholars, first by examining the effect of a wide number of social, situational, attitudinal and perceptual variables, then by developing three multiple regression models, one for the entire sample of scholars, one for a subgroup using primary materials in their work "often" or "occasionally," and one for a subgroup using primary materials "seldom" or "never." The models were also used to test a theory that there is an interaction effect between the norms imposed on information seeking by a discipline or a sub-discipline, the individual scholar's own attitude toward information seeking, and amount of campus library use. The population for the study was over 1,000 literary scholars teaching at 35 Canadian universities. Following interviews and a pre-test, a questionnaire survey was mailed to a random sample of 250 scholars for which a response rate of 64% was achieved. A number of long-held assumptions about the information seeking behaviour of literary scholars were supported by the findings of the study; e.g. that literary scholars are reluctant to delegate searching and that informal communication is not of great importance to them. A number of other assumptions, however, were not supported by the findings: e.g. that browsing is an important way of locating information and that literary scholars are skeptical about the value of using computers in bibliographic searching. The regression models developed for the two sub-groups of scholars were the most successful. The model developed for those scholars who used primary source materials in their work "often" or "occasionally" contained six variables and explained 62% of the variance of campus library use. The model developed for those scholars who used primary source materials "seldom" or "never" contained two variables and explained 44% of the variance of campus library use. The models further revealed that the attitude of the individual scholar toward information seeking was a good predictor of campus library use for those scholars who were judged to be working in areas with weak norms governing information seeking behaviour (those who used primary source materials "seldom" or "never" in their work) and a poor predictor for those scholars working in areas judged to have strong norms governing information seeking behaviour (those who used primary source materials "often" or "occasionally" in their work). AN University Microfilms Order Number ADG88-02459. AU MACIUSZKO, KATHLEEN LYNN. IN Case Western Reserve University Ph.D 1987, 320 pages. TI HARDCOPY VERSUS ONLINE SEARCHING: A STUDY IN RETRIEVAL EFFECTIVENESS. DE Library Science. AB This study examined retrieval effectiveness from the perspective of hardcopy and online searching. The experiment tested the following hypothesis: Hardcopy searching is a more effective way to search by subject for bibliographic information than is online searching. Its primary aim was to determine which mode was more effective for processing queries of the type commonly encountered in a general reference department of an academic or public library. This study made use of two equivalent databases to study two different modes of retrieval, and concentrated on information sources available to the general public as opposed to a specialized audience such as scientists. The methodology involved three steps: creating a set of questions, searching them in hardcopy indexes and their online counterparts, and evaluating the results. Six different college students furnished 22 test questions. Two students formulated 8 questions covering the field of biology. Two others supplied 7 questions in the field of business. Another 2 formulated questions of popular interest. Librarians in academic and public libraries searched the 22 test questions in hardcopy bibliographic indexes and their equivalent online versions for a two-year period. The experiment yielded a total of 88 search results. The following H. W. Wilson indexes and their online counterparts (available through Wilsonline) were used: Biological & Agricultural Index, Business Periodicals Index, and Readers' Guide to Periodical Literature. In general, the findings did not support the assumption that hardcopy searching produced better search results than online searching. It was not a more effective way to search by subject for bibliographic information than was online. The most striking finding of the experiment was that neither mode of searching offered an overall noteworthy advantage over the other. It was interesting to note that there was very little duplication of citations between the modes and that retrieval results were low for both modes. Neither finding was new with regard to online searches. However, placed in the framework of this experiment, they gain added significance because the same results surfaced with respect to the hardcopy searches. Without a doubt, regardless of the mode, the effectiveness level of information retrieval systems is not where it should be. (Abstract shortened with permission of author.). ------------------------------ END OF IRList Digest ********************