Dissertation Information for Carlos Vidal-Arbona
- Carlos Vidal-Arbona
- Library and Information Science
- Case Western Reserve University (USA) (1986)
- Miranda Pao
- Sarah Scott Gibson
- Tefko Saracevic
MPACT Status: Complete - Except Indecipherables
Title: COMPARING THE RETRIEVAL EFFECTIVENESS OF FREE-TEXT AND CITATION SEARCH STRATEGIES IN THE SUBJECT OF TECHNOLOGY PLANNING
Abstract: This comparative study addresses the retrieval effectiveness of two methods of online searching: subject term and citation searching. Both methods were used to search the subject of Technology Planning in a collection of articles published by 223 journals.
The objective of the experiment was to determine: (1) how much overlap existed between the articles retrieved by each of the methods; (2) under what circumstances each method yielded more relevant articles; (3) if a method which consistently retrieves a high percentage of relevant articles can be developed.
Searches were conducted for eight questions on the Dialog Information Service against two data bases: (1) ABI/INFORM--a management literature file containing bibliographic citations and article summaries; (2) Social SciSEARCH--a citation index to the literature of the Social Sciences.
Each of the questions was searched twice: one via subject terms on ABI/INFORM, and the second time via citations on Social Scisearch. Practitioners in Technology Planning evaluated the search results for relevance. Effectiveness scores were calculated based on relevance judgements performed by experts on how similar the retrieved results were to two known relevant "seed" articles they submitted for the experiment.
The study results demonstrated that subject term searches satisfied the information requirement of higher recall with reasonable precision preferred by the experts who participated in the investigation. Citation was found to be an effective method when "seed" documents cited significant prior work. Additionally, it was shown that the number of common documents retrieved by both methods is very small. And that both methods are complementary. This finding is consistent with prior research and implies that subject bibliographies assembled by employing one method alone miss a large proportion of relevant articles.