Search

A computer-based artificial neural network (ANN) that learns to classify sentences in a text as important or unimportant is described. The program is designed to select the sentences that are important enough to be included in composition of an abstract of the text. The ANN is embedded in a conventional symbolic environment consisting of lexical/semantic, morphological, syntactic, and pragmatic analyzers and synthesizers. Only certain features are computed by these symbolic modules, those that are more relevant to this sentence classification task. The selected features are translated into vectors of ones, zeros, and intermediate values, and input into the network. In this way, the ANN collectively considers all linguistic and pragmatic levels in making a decision. This is in contrast to most text abstraction systems, which use information retrieval techniques such as keyword extraction. Preliminary testing shows the ANN has a success rate of 86 percent with 10 different sets of novel sentences, after having been trained on disparate sets of 90 sentences. Contains 12 references. (Author/MSE)

A study (the first of two studies) examined the effects that training in annotation writing had on college students' comprehension and summary writing abilities as compared with students who received a more traditional type of instruction. One half of the 144 university students enrolled in a college reading and study skills course received a treatment which consisted of training in annotation writing. The other half received a more traditional type of college reading instruction. Students completed pretests and posttests measuring comprehension and measuring summary writing. Results indicated that: (1) students in the annotation training group had scores which were statistically significantly greater for length, combined indexes, and efficiency ratios, but not for time, key ideas, comprehension, and peer evaluations. Findings of the first study suggest that students who received annotation training were able to produce succinct summaries. The second study determined the effects that training in annotation writing had on college students' critical and evaluative thinking. The subjects were 100 university students enrolled in a college reading and study skills course. Three of six classes received instruction in annotation writing. The three remaining classes received a more traditional type of college reading and study skills instruction. A critical judgments test was used as a pretest and posttest. Results indicated no statistically significant differences between the experimental and control groups. Findings suggest that annotation training did not appear to improve critical and evaluative thinking skills. (Two figures and one table of data are included.) (RS)

To identify the procedures for selecting documents for inclusion in the Resources in Education (RIE) portion of the ERIC database, the directors of each of the ERIC system's 16 clearinghouses were contacted and asked the following questions: What is the acceptance rate of unsolicited manuscripts submitted to your clearinghouse for possible publication? What are the academic credentials and experience of those who evaluate unsolicited manuscripts? and What criteria are used to evaluate the quality of unsolicited manuscripts? Their replies indicate that: (1) each clearinghouse uses the services of professional experts with advanced degrees and experience in the particular areas of education which it is responsible for covering to review the manuscripts it receives; (2) reviewers follow an explicit and impressive set of criteria to decide which documents to include in the database; (3) the impressive dissemination system of the ERIC database includes microfilming the entire document, abstracts printed in a monthly periodical, the capability for Boolean searches of the database through online and CD-ROM systems, and depositories of the microfiche collection throughout the United States and several other nations; and (4) the ERIC system has demonstrated a commitment to provide useful information to the professional educator in colleges and universities, as well as to the teacher, administrator, and other practitioners in the K-12 school systems. The contribution of faculty from several universities to RIE is analyzed and it is concluded that Wayne State College faculty have not contributed their fair share of writing and scholarship to the ERIC database. (SD)

Discusses issues related to natural language processing, including theoretical developments; natural language understanding; tools and techniques; natural language text processing systems; abstracting; information extraction; information retrieval; interfaces; software; Internet, Web, and digital library applications; machine translation for multilingual retrieval; and evaluation. (Contains 136 references.) (LRW)

The writing of academic abstracts is more than a tiresome necessity of scholarly life. It is a practice which goes beyond genre and technique to questions of identity and the promotional economies of academic work. In this paper, a series of abstracts from a variety of refereed journals and conferences are deconstructed and a set of questions that allow individuals to "read" the representation of data, argument, methodology, and significance are developed. According to the paper, to begin to make sense of abstract writing, a selection of different educational journals were surveyed and a rubric was constructed to use to read each item; a second foray into journal abstracts allowed researchers to read for depth rather than breadth and in publications across three countries. The paper shows three abstracts to illustrate the problem of writing anticipatory texts that overpromise, particularly when they are based on research that is hardly yet begun. It argues that the rules of abstract engagement are fluid and increasingly important with the advent of online journals and global citation indices. The paper suggests that abstract art is now an obligatory aspect of postgraduate supervision. It further suggests that the writing of abstracts is neglected in the academy and that there is benefit in being more systematic about supporting postgraduate and newly qualified researchers in writing abstracts. An appendix offers questions that encourage writers to engage with the text work and identity work involved in writing an abstract. (Author/NKA)

Presents various levels of criteria for judging the quality of abstracts and abstracting. Requirements for abstracts to be read by humans are compared with requirements for those to be searched by computer. Concludes that the wide availability of complete text in electronic form does not reduce the value of abstracts for information retrieval. Contains 43 references. (Author/MES)

Indexing and abstraction services in the social sciences are growing as primary sources. Results are that users become overwhelmed with the number of secondary services they use. Therefore, planning, coverage and overlap are examined in this research report with emphasis on reexamining secondary services and redefining users and users' needs. Overlap and coverage of secondary sources dealing with criminology and public administration were examined. Results showed that adequate but insufficient services are provided. Recommendations of this study are that the decision must be made whether broad or narrow coverage is to occur. It suggests that an identification be made of the material and user evaluation of services. (AP)

Established in 1968, The European Documentation and Information System for Education (EUDISED) has passed through several developmental stages. Initially, early attempts to apply computer techniques to educational documentation and information were not coordinated between European countries. Interest in the American ERIC system resulted in two studies. The first study (1970) concerned the problems of using a common thesaurus of descriptors and a common format. The second study (1971) designed a proposed system. In 1974, the multilingual EUDISED Thesaurus was published and field experiments began to test the proposed system. In the primary experiment twelve national agencies completed a common worksheet on selected R & D projects in their countries, and the results were published in the EUDISED R & D Bulletin. This and other experiments led to increased standardization between countries regarding indexing, abstracting, and dissemination. Present concerns are for the maintenance of a self-sustaining system; the expansion of computer access to all agencies; and the access to ERIC material and data from various international organizations. (DAG)

The purpose of this project was to provide guidance to ERIC on how to facilitate product information retrieval. Information on practitioner needs, on the use of descriptors and abstracting procedures, and on alternatives for improving the system was gathered and circulated to NIE staff members. The resulting document described three alternative ways of improving access to practitioner oriented documents and presented draft guidelines for writing abstracts. The alternatives were: revising the method of descriptor assignment to produce documents; revising and expanding pubtype codes; and creating a separate system for all products, programs, and practice documents. This document was then sent out to all clearinghouse directors, directors of ERIC search services, and others with a request for comments. The responses were summarized, sample abstracts were developed and six recommendations were made: (1) pubtype codes should be refined, pilot tested, and refined again; (2) inter-clearinghouse inconsistency in indexing and abstracting should be reduced; (3) guidelines for abstracting product information should be agreed upon and entered into the Thesaurus; (4) successful computer search strategies for retrieving product documents should be available to all; (5) existing product information systems should be considered for coordination with ERIC. The above documents are presented in this report along with the responses. (DAG)

Six major indexing and abstracting services were studied to isolate similarities and differences, obtain raw data, and attempt to estimate primary material covered by the services. The 1973 issues of all services were examined and statistics were obtained for periodical and non-periodical entries. Data were analyzed and the services compared by: (1) subject arrangement, (2) coverage and overlap of periodicals, (3) core periodicals, (4) coverage of nonjournal material and archives, (5) timeliness, (6) bibliographic description, and (7) origin of abstracts. Services were found to be uneven in coverage and timeliness. The authors concluded that: (1) each service's view of the scope of the field governs coverage of library science and/or information science literature, and periodical and/or nonperiodical literature, and (2) the low circulation of the services may hinder financing of improvements. It was recommended that the editors and representatives of the publishers meet to explore the possibility of linking English language services into a single, comprehensive service. Extensive data tables, lists, and service profiles are included. (KP)

The Design of Information Systems in the Social Sciences project (DISISS) of the University of Bath, England, has recently completed evaluations of three programs: (1) a library staff digest intended to inform personnel of the inner workings of the library; (2) an information system designed to assist professionals and private individuals working with young people; and (3) an indexing and abstracting service--Geo Abstracts. The assessment is done largely through the use of surveys, but cost-benefit analysis and use studies are discussed as tools to further refine the evaluation process. (EMH)

A workshop was developed at the Bank of Canada to give instruction in writing brief summaries of financial analyses to junior economists entering the bank after university. These employees were expected to write these analyses for the senior officers of the institution. It had been found that the specialists had not learned strategies for exploring their statistical data with the objective of identifying the deeper meaning or story in the data and then presenting it in the expected format. They had been following a restrictive sequence of procedures and producing unconnected series of relatively superficial observations about their data, which did not satisfy their audience's need for a story. The workshop had three goals: to make explicit the information needs of the readers, to develop the specialists' ability to use writing strategies for better analysis, and to enable the specialists to provide sharply focused and effectively structured summaries that represented their best judgment, as technical experts, about the essential story contained in the data. The content of the workshop included discussion of the writing process, direct contact with the audience for the texts, and instruction in developing a preliminary writing plan, drafting, and revision. The workshop has resulted in texts containing more meaningful analyses, more direct discussion of the analyses between the specialists and the senior officer, and increased confidence and analytical capabilities in the specialists. (MSE)

The purpose of this study was to develop a model for the distribution of authorships--based on the initial hypothesis that the distribution of authorships follows a shifted Waring distribution--and to test the derived model and some other discrete probability models for goodness-of-fit against empirical data. Bibliographic data from 15 abstracting journals covering the literature in six fields--engineering, medical, physical, mathematical and social sciences, and humanities--were used in testing the goodness-of-fit of the shifted Waring distribution and 13 other discrete probability models. The preliminary findings presented here are based on 60 data sets collected from 10 abstracting journals covering the literature in the mathematical and social sciences and humanities. They indicate that the promising models for the distribution of authorships are the shifted Waring, shifted generalized negative binomial, shifted negative binomial, shifted generalized Poisson, and shifted inverse Gaussian-Poisson distributions. Three advantages and possible practical applications of a model for the distribution of authorships include: (1) ability to summarize the entire frequency distribution by a few parameters of the model; (2) estimation of the number of entries in an author index; and (3) usefulness in a simulation study designed to determine, subject to space constraints, the maximum number of authors per paper to be included in an author index. Forms of the discrete probability models are appended. (5 tables, 12 references) (Author/CGD)

Twenty-two reports of research focusing on the composing processes of college students of English as a Second Language are reviewed. The research, reported between 1982 and 1989, consists of case studies of various issues: (1) general accounts of the second language composing process; (2) revision; (3) attention to rhetorical form; (4) monitor use; (5) abstracting; (6) reader awareness; and (7) planning. The major findings of the research are discussed, and particular reference is made to the most frequently and/or extensively addressed issues: recursion, variation in the second language composing process, planning, transcribing, revising, first language use in second language composition, audience awareness, monitoring, and the transfer of first language writing skills to second language composition. The studies' methods, reporting, and interpretation practices are also described and critiqued. It is concluded that while the studies' findings are interesting, provocative, and potentially useful, their credibility is not beyond question. (MSE)

The origin and continuation of the futures research seminar at the School of Education of the University of North Carolina at Chapel Hill are discussed. The continuing research seminar takes place throughout the calendar year for students who wish to use futures research methods in their dissertations. One of the major projects of the seminar is the development of an environmental scanning database. Each student is assigned specific journals and newspapers to scan in order to augment the data gathered by students. Each student is also assigned one or more professional journal(s) to search for pertinent literature dealing with a number of categories describing the methodological and substantive scope of the seminar. Each student is assigned categories of the literature database to maintain. This responsibility includes abstracting articles in that category for insertion in the electronic literature database. The software support systems and the ED QUEST planning model that can be employed by students are briefly considered. Included are: the environmental scanning taxonomy, a list of seminar information resources, criteria and instructions for abstracting, a list of journals scanned for futures research database, and literature database categories. 14 references. (SW)

A plan to establish an environmental scanning and forecasting system for colleges and universities is discussed as a way to maximize long-range planning. After proposing a program structure, attention is directed to methods of gaining organizational acceptance, developing a comprehensive taxonomy with an electronic filing system, identifying and assigning information resources, securing scanners, and training scanners and abstracters. Brief descriptions are included of activities involving the scanning committees, including conduct of meetings, training committee members in forecasting techniques, developing the scanning newsletter, and writing issue briefs. The primary purpose of an environmental scanning and forecasting system is to identify and analyze trends in order to uncover emerging issues that may affect the ability of colleges to perform their missions. The desired result would also be to provide the college's senior management with complete, objective, and detailed information about the external environment to help anticipate what is happening in the state, region, and nation. In support of the college's strategic planning process, the scanning committee would analyze and evaluate scanning abstracts on a monthly or bimonthly basis. (SW)

This report presents results from an internal evaluation conducted between September 1981 and August 1982 of abstracts published by the Educational Resources Information Center (ERIC) system. A brief introduction describes the study methodology and reviews limitations, indicating that 526 evaluations of 349 abstracts were conducted by personnel at each of the 16 ERIC Clearinghouses. Next, results of the evaluations prepared by each Clearinghouse are presented, with Clearinghouses being identified only by a random letter of the alphabet. For each Clearinghouse, information is provided on the numbers of abstracts evaluated, evaluations completed, abstractors whose work was represented, and evaluators; overall ratings; types of abstracts represented (i.e., informative, indicative, or mixed) and analysis of their appropriateness; evaluations of abstract style, vocabulary, and content; length of the abstracts evaluated; and details on how the Clearinghouse conducted the evaluations. Next, outcomes are described for the system as a whole, indicating that 67 excellent ratings, 188 good ratings, 150 fair ratings, 91 marginal ratings, and 30 poor ratings were assigned. Common problems included inappropriate selection of abstract type and the omission of important sections of documents. Recommendations for improving abstracts are included. An interim report from an initial pilot study; revisions of the abstract evaluation form; feedback from Clearinghouses regarding the form; sample evaluations; and materials from an abstracting workshop are appended. (BCY)

Discusses research dealing with the cognitive aspects of formal systems of knowledge representation. Highlights include the origins and theoretical foundations of the cognitive viewpoint; cognition and information science; cognitivism, mentalism, and subjective individualism; categorization; mental models; and sociocognitive approaches to indexing and abstracting. (Contains 170 references.) (Author/LRW)