DE eng

Search in the Catalogues and Directories

Hits 1 – 2 of 2

1
Understanding the effects of negative (and positive) pointwise mutual information on word vectors
Salle, A.; Villavicencio, A.. - : Taylor & Francis, 2022
Abstract: Despite the recent popularity of contextual word embeddings, static word embeddings still dominate lexical semantic tasks, making their study of continued relevance. A widely adopted family of such static word embeddings is derived by explicitly factorizing the Pointwise Mutual Information (PMI) weighting of the cooccurrence matrix. As unobserved cooccurrences lead PMI to negative infinity, a common workaround is to clip negative PMI at 0. However, it is unclear what information is lost by collapsing negative PMI values to 0. To answer this question, we isolate and study the effects of negative (and positive) PMI on the semantics and geometry of models adopting factorization of different PMI matrices. Word and sentence-level evaluations show that only accounting for positive PMI in the factorization strongly captures both semantics and syntax, whereas using only negative PMI captures little of semantics but a surprising amount of syntactic information. Results also reveal that incorporating negative PMI induces stronger rank invariance of vector norms and direction, as well as improved rare word representations.
URL: https://eprints.whiterose.ac.uk/184560/
BASE
Hide details
2
Improving tokenisation by alternative treatment of spaces
BASE
Show details

Catalogues
0
0
0
0
0
0
0
Bibliographies
0
0
0
0
0
0
0
0
0
Linked Open Data catalogues
0
Online resources
0
0
0
0
Open access documents
2
0
0
0
0
© 2013 - 2024 Lin|gu|is|tik | Imprint | Privacy Policy | Datenschutzeinstellungen ändern