Abstract—The semantic similarity between two words can be determined based on their common and distinctive features after transforming them into measurable values. Up to now, a variety of transfer functions with respect to the transformation and the combination of features have been proposed. However, none of them have ever processed and combined those features properly, thus making them incapable of making a judgment of similarity that is close to human judgments of similarity. This paper offers a method that represents the optimal combination of the optimal transformation of two types of widely used features. The results obtained from a standard data set show that the proposed solution outperforms all of its benchmarks significantly.
Index Terms—Semantic similarity, transfer function, word similarity.
Autors are with the Faculty of Computer Science and Information Technology, Selangor, Malaysia (e-mail: hochukfong@yahoo.com; masrah@fsktm.upm.edu.my; rabiah@fsktm.upm.edu.my; shyamala@fsktm.upm.edu.my).
Cite: ChukFong Ho, Masrah Azrifah Azmi-Murad, Shyamala Doraisamy, and Rabiah Abdul-Kadir, "Measuring Word Similarity Based on the Optimal Transformation and Combination of Features," International Journal of Machine Learning and Computing vol. 2, no. 3, pp.188-194, 2012.