Word discrimination and sense clustering
Research Field:Computer Science
Lead PI:Dr Martin Emms
Abstract:Within the field of Natural Language Processing (NLP), word discrimination and sense clustering is a type of unsupervised word sense disambiguation technique that aims to distinguish automatically the sense of words by clustering their contexts. Word discrimination has many applications such as Machine Translation, Word Alignment, Information Retrieval, among others. In this particular research we will apply word discrimination methods in order to improve word alignment scores and also to track semantic change of words in time-annotated text corpora, as well as semantic variation of words across corpora written in different dialects of the same language (e.g. Irish English vs American English).