Navigating word association norms to extract semantic information

Abstract

We present a simple model that allows the extraction of semantic similarity relations from free association information. In our study, we use two acclaimed databases of linguistic relationships between pairs of words, feature-based and association-based. We apply a complex networks methodology to disentangle feature based relationships on top of a free association network. As a consequence, we broaden complex networks' applications in the field of psycholinguistics, from a merely descriptive to a predictive level. Results are systematically compared to those of two powerful well-known computational models (LSA and WAS).


Back to Saturday Posters