Article published in:How to do Linguistics with R: Data exploration and statistical analysis
[Not in series 195] 2015
► pp. 323–332
Chapter 16. Introduction to Semantic Vector Spaces
Cosine as a measure of semantic similarity
This chapter introduces Semantic Vector Spaces, another distributional approach to semantics. This method originates in Natural Language Processing. Unlike Behavioural Profiles discussed in the previous chapter, it uses automatically extracted co-occurrences of target words and contextual features. The characteristic features of the method are weighted co-occurrence frequencies and the use of the cosine as the most popular similarity measure. This chapter provides a general introduction to the method, with a case study of English cooking verbs as an illustration.