Word-Space Approach To Case-Based Retrieval

Abstract

This paper presents a strategy for medical articles retrieval, a process also known as case-based retrieval. The proposed strategy is a word-space approach and uses Terrier IR search engine for indexing and retrieval. The medical articles were indexed on different parts of their content to determine which part provides best performance in indexing and retrieval. In the retrieval phase, experiments with six different weighting models were performed. The results show that the best performance is achieved by indexing the whole content of the articles (including the image captions) and by using the BM25 weighting model in the retrieval phase.