Determining domain-specific differences of polysemous words using context information

In this paper we describe our early work with determining domain-specific differences of polysemous words. The goal is to determine whether a word has different meanings in two different text corpora. Previous work tried to identify such words by training word embeddings specific for those domains and then comparing the resulting vectors. Our use case, however, does not contain enough text to train word embeddings. Thus, we present a novel method that is able to make use of context information and determine domain-specific differences of words without generating word embeddings.