An experiment in automatic hierarchical document classification

doi:10.1016/0306-4573(83)90064-X

Information Processing & Management

Volume 19, Issue 3, 1983, Pages 113-120

https://doi.org/10.1016/0306-4573(83)90064-X Get rights and content

Abstract

A method of automatic document classification was developed as part of a larger research project in materials selection. Documents classed as QA by the Library of Congress classification system were clustered at six thresholds by keyword using the single link technique. The automatically generated clusters were then compared to the Library of Congress subclasses to which the documents had been assigned by human classifiers. Finally, a partial classified hierarchy was formed from the individual document clusters within a single threshold. Implications of the utility of grouping documents for on-line searching are discussed.

References (12)

L.B. Doyle
Breaking the Cost Barrier in Automatic Classification
L.B. Doyle
Breaking the Cost Barrier in Automatic Classification
H. Borko
Measuring the reliability of subject classification by men and machines
Am. Documentation
(1964)
N.S. Prywes
Browsing in an automated library through remote access
L.B. Doyle
Breaking the Cost Barrier in Automatic Classification
(1966)
K.L. Kwok
Cited titles: a new source of keyword extraction for automatic document classification and retrieval
Information Utilities: Proceedings of the 37th ASIS Annual Meeting
(1974)

There are more references available in the full text version of this article.

Cited by (15)

Optimizing SCImago Journal & Country Rank classification by community detection
2014, Journal of Informetrics
Citation Excerpt :
Another topic commonly addressed by the scientific literature on classification is the adequacy and possibility of developing automatic classification systems to avoid, as far as possible, human intervention. Early works were developed by authors such as Luhn (1957) in Information Retrieval scope at the end of 1950s; but interest remained strong in the 1960s (Garland, 1982), and furthered with the advance and development of scientific databases, bibliometric indicators, science mapping, etc. up to the present. Some research reviewed here tried to avoid human intervention, but concluded it was not possible to do so completely.
Subject classification arises as an important topic for bibliometrics and scientometrics, searching to develop reliable and consistent tools and outputs. Such objectives also call for a well delimited underlying subject classification scheme that adequately reflects scientific fields. Within the broad ensemble of classification techniques, clustering analysis is one of the most successful.
Two clustering algorithms based on modularity – the VOS and Louvain methods – are presented here for the purpose of updating and optimizing the journal classification of the SCImago Journal & Country Rank (SJR) platform. We used network analysis and Pajek visualization software to run both algorithms on a network of more than 18,000 SJR journals combining three citation-based measures of direct citation, co-citation and bibliographic coupling. The set of clusters obtained was termed through category labels assigned to SJR journals and significant words from journal titles.
Despite the fact that both algorithms exhibited slight differences in performance, the results show a similar behaviour in grouping journals. Consequently, they are deemed to be appropriate solutions for classification purposes. The two newly generated algorithm-based classifications were compared to other bibliometric classification systems, including the original SJR and WoS Subject Categories, in order to validate their consistency, adequacy and accuracy. In addition to some noteworthy differences, we found a certain coherence and homogeneity among the four classification systems analysed.
Is the lack of keyword synergism inhibiting maturation in the mis theory? An exploratory perspective
1998, Information Processing and Management
This article explores previous research on keyword search techniques and evaluates the comprehensiveness of a well-known keyword classification scheme. Comparison of literature reviews provide evidence that this topic is not well agreed upon and these contradictions have reduced reliance on keyword searches as a method of exploring prior research on a given topic. Presently, there does not exist a comprehensive management information systems (MIS) keyword list. Information regarding keyword usage in 14676 articles published in 27 MIS journals from 1985–1994 was collected electronically from an ABI Inform database. After several levels of refinements on 4386 keywords, a comprehensive list of 1056 keywords was obtained. Our MIS keyword list was compared with that of Barki and differences with Barki's list were identified, such as: (i) failing to include many keyword subjects that have frequently occurred during the past decade, and (ii) inclusion of keywords that have not had a single occurrence in MIS journals for the same period.
The interpretations of library use in the age of digital libraries: Virtualizing the name
1997, Library and Information Science Research
There have been numerous changes in the way that libraries organize information and how users seek, gain access to, and obtain library materials. These changes have modified many traditional library services, introduced new jargon, and created new library uses. If students are to become self-sustaining and effective users of information sources and services, they need to understand various library uses. This article reports a case study designed to examine the extent to which college students agree in interpreting the term library use. The degree of variation in how these students interpret basic library activities is high. The study's findings suggest that the notion of the library as a space is better agreed upon than the notion of the library as a store or service. Since the library use study is carried out in the context of rapidly changing information technology, this two-prong article will first discuss five different shifts that characterize the transitional nature of today's libraries; then the library use study will be discussed, and possible conclusions will be drawn. The findings have wide implications for designing user-centered information access instructional programs.
Recent trends in hierarchic document clustering: A critical review
1988, Information Processing and Management
This article reviews recent research into the use of hierarchic agglomerative clustering methods for document retrieval. After an introduction to the calculation of interdocument similarities and to clustering methods that are appropriate for document clustering, the article discusses algorithms that can be used to allow the implementation of these methods on databases of nontrivial size. The validation of document hierarchies is described using tests based on the theory of random graphs and on empirical characteristics of document collections that are to be clustered. A range of search strategies is available for retrieval from document hierarchies and the results are presented of a series of research projects that have used these strategies to search the clusters resulting from several different types of hierarchic agglomerative clustering method. It is suggested that the complete linkage method is probably the most effective method in terms of retrieval performance; however, it is also difficult to implement in an efficient manner. Other applications of document clustering techniques are discussed briefly; experimental evidence suggests that nearest neighbor clusters, possibly represented as a network model, provide a reasonably efficient and effective means of including interdocument similarity information in document retrieval systems.
Confidence Range: Bridging Failure Detection and True Class Probability on Selective Hierarchical Text Classification
2022, SSRN
Library applications of knowledge-based systems
2019, Expert Systems in Reference Services

View all citing articles on Scopus

View full text

An experiment in automatic hierarchical document classification

Abstract

Breaking the Cost Barrier in Automatic Classification

Breaking the Cost Barrier in Automatic Classification

Measuring the reliability of subject classification by men and machines

Am. Documentation

Browsing in an automated library through remote access

Breaking the Cost Barrier in Automatic Classification

Cited titles: a new source of keyword extraction for automatic document classification and retrieval

Information Utilities: Proceedings of the 37th ASIS Annual Meeting