Approximating Hierarchy-Based Similarity for WordNet Nominal Synsets using Topic Signatures

Topic signatures are context vectors built for concepts. They can be automatically acquired for any concept hierarchy using simple methods. This paper explores the correlation between a distributional-based semantic similarity based on topic signatures and several hierarchy-based similarities. We show that topic signatures can be used to approximate link distance in WordNet (0.88 correlation), which allows for various applications, e.g. classifying new concepts in existing hierarchies. We have evaluated two methods for building topic sigantures (monosemous relatives vs. all relatives) and explore a number of different parameters for both methods.