Abstract—Automatically identifying the intended sense of ambiguous words improves the performance of clinical and biomedical applications. This paper by proposing Optimized Gloss Vector relatedness and Adapted Gloss Vector similarity measures, two enhanced semantic measures based on Gloss Vector relatedness measure (GV), evaluates their effectiveness over the task of word sense disambiguation (WSD) in the biomedical domain. Generally, GV measure, after computation of the concepts’ gloss vectors using their definitions and an external corpus, quantifies the degree of relatedness as the cosine of the angle between two input concepts’ computed gloss vectors. We use Pointwise Mutual Information (PMI) and Medical Subject Heading (MeSH) Structure for GV optimization and similarity adaptation respectively. The experimental result on the WSD dataset shows the proposed definition-based measures outperform other semantic measures in terms of accuracy.

Index Terms—Word sense disambiguation, PMI, semantic relatedness, semantic similarity, MeSH, biomedical text mining, bioinformatics.

Ahmad Pesaranghader is with the Faculty of Creative Multimedia, Multimedia University, Jalan Multimedia 63100, Cyberjaya, Selangor, Malaysia (e-mail: ahmad.pgh@sfmd.ir).
Ali Pesaranghader and Norwati Mustapha are with the Faculty of Computer Science and Infromation Technology, Universiti Putra Malaysia, 43400 UPM, Serdang, Malaysia (e-mail: ali.pgh@sfmd.ir, norwati@upm.edu.my).

Cite: Ahmad Pesaranghader, Ali Pesaranghader, and Norwati Mustapha, "Word Sense Disambiguation for Biomedical Text Mining Using Definition-Based Semantic Relatedness and Similarity Measures," International Journal of Bioscience, Biochemistry and Bioinformatics vol. 4, no. 4, pp. 280-283, 2014.