← smac.pub home

A Deeper Look into Dependency-Based Word Embeddings

pdf bibtex data slides poster 20 citations workshop paper

Authors: Sean MacAvaney, Amir Zeldes

Appeared in: Proceedings of the NAACL-HLT 2018 Student Research Workshop (NAACL Student Workshop 2018)

DOI 10.18653/v1/N18-4006 DBLP conf/naacl/MacAvaneyZ18 ACL N18-4006 arXiv 1804.05972 Google Scholar 7wWfoDgAAAAJ:UeHWp8X0CEIC Semantic Scholar b13b153cd70492c0b821bf378a4b6b7f2cfe0c58 smac.pub naaclsrw2018-depwembed


We investigate the effect of various dependency-based word embeddings on distinguishing between functional and domain similarity, word similarity rankings, and two downstream tasks in English. Variations include word embeddings trained using context windows from Stanford and Universal dependencies at several levels of enhancement (ranging from unlabeled, to Enhanced++ dependencies). Results are compared to basic linear contexts and evaluated on several datasets. We found that embeddings trained with Universal and Stanford dependency contexts excel at different tasks, and that enhanced dependencies often improve performance.

BibTeX @inproceedings{macavaney:naaclsrw2018-depwembed, author = {MacAvaney, Sean and Zeldes, Amir}, title = {A Deeper Look into Dependency-Based Word Embeddings}, booktitle = {Proceedings of the NAACL-HLT 2018 Student Research Workshop}, year = {2018}, url = {https://arxiv.org/abs/1804.05972}, doi = {10.18653/v1/N18-4006}, pages = {40--45} }