Taxonomy Learning using Term Specificity and Similarity

Learning taxonomy for technical terms is difficult and tedious task, especially when new terms should be included. The goal of this paper is to assign taxonomic relations among technical terms. We propose new approach to the problem that relies on term specificity and similarity measures. Term specificity and similarity are necessary conditions for taxonomy learning, because highly specific terms tend to locate in deep levels and semantically similar terms are close to each other in taxonomy. We analyzed various features used in previous researches in view of term specificity and similarity, and applied optimal features for term specificity and similarity to our method.