Ira Leviant

Research

I am interested in a variety of topics in NLP, computational linguistics, computational language acquisition and machine learning. My main research area is multilingual vector space models.

Data

MultilingualWS353: MultilingualWS353 is a translated dataset of the popular wordsim353 dataset to the following three languages: German, Italian and Russian. The translated dataset which consists of 350 word pairs in each langauge was annotated for similarity by human evaluators through the CrowdFlower crowdsourcing service based on annotation guidelines in the corresponding language.

MultilingualSimLex999: MultilingualSimLex999 is a translated dataset of the popular SimLex999 dataset to the following three languages: German, Italian and Russian. The translated dataset which consists of 999 word pairs in each langauge was annotated for similarity by human evaluators through the CrowdFlower crowdsourcing service based on annotation guidelines in the corresponding language.

Publications

• Ira Leviant and Roi Reichart. “Iterative Word Embeddings”,  in submission to EACL 2021.

• Ivan Vulic, Simon Baker, Edoardo Maria Ponti, Ulla Petti, Ira Leviant, Kelly Wing, Olga Majewska, Eden Bar, Matt Malone, Thierry Poibeau, Roi Reichart, and Anna Korhonen.  “Multi-simlex: A largescale evaluation of multilingual and cross-lingual lexical semantic similarity”, Computational Linguistics, 2020.

• Nikola Mrkšic, Ivan Vulic, Diarmuid Ó Séaghdha, Ira Leviant, Roi Reichart, Milica Gašic´,  Anna Korhonen and Steve Young. “Semantic specialization of distributional word vector spaces using monolingual and cross-lingual constraints”, Transactions of the Association of Computational Linguistics, s, 5(1):309–324, 2017.

• Ira Leviant and Roi Reichart. “Separated by an Un-common Language: Towards Judgment Language Informed Vector Space Modeling”,  Preprint published on arXiv. arxiv:1508.00106, 2015.