CompareWords: Measuring semantic change in word usage in different corpora
Software Impacts (2021)
Abstract
We present CompareWords; A software package developed for measuring semantic change of particular words between two corpora. We have used it for measuring changes in meaning between two time periods, but it could also be used to measure changes in meaning between different topic areas or literary genres. Our technique uses word-embeddings for each corpus, and cross-lingual transformations. Thus it requires the corpora to be large enough to train good word-embeddings.
Authors
BibTex
@article{TAYLOR2021100067, title = {CompareWords: Measuring semantic change in word usage in different corpora}, journal = {Software Impacts}, volume = {8}, pages = {100067}, year = {2021}, issn = {2665-9638}, doi = {https://doi.org/10.1016/j.simpa.2021.100067}, url = {https://www.sciencedirect.com/science/article/pii/S2665963821000154}, author = {Stephen Taylor and Pavel Přibáň and Ondřej Pražák}, keywords = {Orthogonal transformation, Canonical correlation analysis, Word-embeddings, Lexical semantic change detection}, abstract = {We present CompareWords; A software package developed for measuring semantic change of particular words between two corpora. We have used it for measuring changes in meaning between two time periods, but it could also be used to measure changes in meaning between different topic areas or literary genres. Our technique uses word-embeddings for each corpus, and cross-lingual transformations. Thus it requires the corpora to be large enough to train good word-embeddings.} }
