Using Thematic Contexts and Previous Solutions for Maintaining and Accessing Institutional Repositories

Abstract

Institutional Repositories are collections of digital resources organized to facilitate their long term access. Two fundamental issues that need to be addressed at the moment of implementing these repositories are their maintenance and search instruments. This article describes an intelligent tool whose goal is to overcome certain general limitations encountered in current mechanisms for managing Institutional Repositories. The novel aspects of the proposed tool are the use of previous examples to generate suggestions for cataloging incoming digital resources and the ability to refect a thematic context during the search process. The proposed tool applies casebased reasoning in combination with information retrieval techniques to take advantage of a large set of previously classified resources with the purpose of supporting the cataloger's task. The tool has been evaluated using a collection consisting of 9792 training examples from the SciELO electronic library and a test set containing 100 articles coming from diferent disciplines.