Year: 2018
Type: preprint
Abstract: We present a new method for estimating vector space representations of words: embedding learning by concept induction. We test this method on a highly parallel corpus and learn semantic representations of words in 1259 different languages in a single common space. An extensive experimental evaluation on crosslingual word similarity and sentiment analysis indicates that concept-based multilingual embedding learning performs better than previous approaches.
Source: arXiv (Cornell University)
Authors: Dufter, Philipp, Zhao, Mengjie, Schmitt, Martin, Fraser, Alexander, Schütze, Hinrich
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2953242939
