Knowledge Discovery in Bibliographic Databases

Citation

Limb, P. (2000), "Knowledge Discovery in Bibliographic Databases", Online Information Review, Vol. 24 No. 5, pp. 401-411. https://doi.org/10.1108/oir.2000.24.5.401.5

This issue of Library Trends (as often the case with this excellent journal) presents a significant collection of cutting‐edge library science research. This is not the first special issue on the themes to appear in a professional journal, but it goes beyond a discussion of terminology and methods, and applies the subject directly to areas relevant to library and information science.

Knowledge discovery in databases (KDD) is a multi‐ and inter‐disciplinary field that investigates databases and evaluates data in a search for relevant relationships that might yield new knowledge but which are hidden. In other words data are logically but not bibliographically linked. Data mining identifies new patterns involved in this search. Librarians have been doing just this, under different names such as classification and information retrieval systems, for many years, but this compilation of 13 articles brings together library science, computer science and information science researchers in a fruitful exchange of state‐of‐the‐art views. Contributions range across information technology, classification, citation analysis and abstracting.

Jay Norton provides a concise overview. Barbara Kwasnik then surveys the links between classification (hierarchies, trees, paradigms and faceted analysis) and knowledge. Don Swanson and Neil Smalheiser, initiators of a major project to apply the principles of KDD in biomedicine, present their findings on implicit text linkages and show how the Arrowsmith programme (http://kiwi.uchicago.edu) can create juxtapositions of Medline search records that suggest new relationships. Kenneth Cory (in an article first published in 1997) then successfully applies the Swanson principles to online humanities databases (in this case, humanities index), a more difficult proposition given the less technical language and goals of humanities literature. In similar analogical fashion Henry Small analyses the crossing of disciplinary boundaries in knowledge discovery, while Jian Qin examines the discovering of semantic patterns in bibliographically coupled documents using keyword density. Bipin Desai and others describe CINDI, a virtual library indexing and discovery system that uses metadata in the form of a semantic header containing data on syntactic and semantic content of a document to simulate the expertise of both cataloging and reference librarians. Other articles discuss knowledge discovery in co‐occurrence frequency of pairs of words (Qin He), frequent word sequences (Helena Ahonen), automatic template mining from digital documents and its use in search engines (Gobinda Chowdhury), and criteria for judging the quality of abstracts (Maria Pinto and F.W. Lancaster). Finally, Lixin Yu shows the power of algorithm‐based spatial cartographic information retrieval and argues that MARC records allow automatic conversion of map records from OPACs.

Emeritus Professor Herbert White concludes the issue by asking, with regard to librarian‐IT relations, “which is the tail and which is the dog?” He doubts that computers can be programmed to “sift the chaff from the wheat” effectively; rather, the best knowledge workers in the twenty‐first century are likely to be “qualified information intermediaries [working] on behalf of, and to protect, the end user”. The articles are well edited to comprise a coherent whole, focusing on KDD across different disciplines and methods. They are informative and tackle central issues from innovative angles. There is much here to chew on, from database design to researcher‐librarian interfaces, to future pathways for librarians, and for this reason the work is highly recommended.