COMPARATIVE ANALYSIS OF RANKING FUNCTIONS FOR RETRIEVING INFORMATION FROM MEDICAL REPOSITORY

Abstract

Current and forthcoming Information Retrieval algorithms demand high mean average precision with contemporary high recall rates in the technical literature. Nevertheless, the existing state-of-the-art is still not optimized for speed, average query latency, and performance. The previous researchers presented various information retrieval models in the literature but the user search led to a ranking of documents that were hopeful to be relevant. In this paper, an evaluation of various information retrieval models is presented with a range of algorithms. The aim is to elaborate and review the current information retrieval function in the context of enterprise domain- specific search. Experiments were conducted on the OHSUMED benchmark data set from MEDLINE, a medical information database. The experimental results demonstrate that BM25F ranking function outperforms other extensively used ranking functions such as BM25, TFIDF, and Cosine on precision and recall measures.