The INSPEC Document Database and Thesaurus

INSPEC is the property of the Institute of Electrical and Electronic Engineers - we acknowledge with thanks permission to use a portion of this database for our research.

It comprises records of papers from scientific journals covering (A) Physics, (B) Electrical and Electronic Engineering, (C) Computer Technology and (D) Computer Applications. It is available commercially on CD-ROM, and via the on-line host systems. The section used for the CILKS project consisted of 314427 records under categories (C) and (D), published between October 1989 and September 1993.

See a typical INSPEC document record.

The INSPEC thesaurus is published both in printed form and as an adjunct to the document database, where it is used for query expansion in conventional boolean searching. New issues are brought out every three years or so, to cope with ongoing changes in technical terminology. The (1992) version used for CILKS contains 13477 terms (6454 preferred and 7023 lead-in). Unusually for a standard thesaurus, one lead-in may refer to two or more preferred terms: a point to be taken into account when designing the navigation software.

See a page of typical INSPEC thesaurus term entries.

10953 associative relationships are represented within the thesaurus, and 6184 hierarchical relationships. One term may have two or more "broader" terms, so the hierarchies could be said to form a semi-lattice. As well as a list of term entries in alphabetical order, the printed thesaurus contains hierarchical displays under 552 top terms - these can be generated directly from the CILKS thesaurus database.

See a typical hierarchical display.

INSPEC documents are also indexed with one or more class-codes from an abstract hierarchical classification scheme. Thesaurus terms and class-codes are associated in a many-to-many relationship - again this is unusual for a standard thesaurus since the more normal practice is to associate one term with one class-code. In principle the class-codes represent alternative "paths" into and through the thesaurus which may be exploited for query enhancement. An MSc. project to investigate the effectiveness of this strategy was undertaken in conjunction with the CILKS research (see publications, but the results did not warrant its inclusion in the final system.