SemEval-2017 Task 2: Monolingual and Cross-lingual Word Similarity
This task provides a challenging benchmark for the evaluation of semantic similarity techniques, and in particular for word embeddings. Monolingual data is available for five different languages: English, Farsi, German, Italian, and Spanish. There is an additional subtask on cross-lingual word similarity. This subtask is composed by ten datasets for evaluating ten different pairs of languages, intended for testing cross-lingual semantic similarity techniques as well as bilingual and multilingual word representations. The evaluation datasets provide a well-balanced set of word pairs, including domain specific words, multiwords and named entities.
If you are interested, please visit our website to find more details or to register and download the trial data: http://alt.qcri.org/semeval2017/task2/
Please note that the evaluation takes place on January 2017.
ORGANIZERS
Jose Camacho Collados, Sapienza University of Rome, Italy
Mohammad Taher Pilehvar, Cambridge University, UK
Nigel Collier, Cambridge University, UK
Roberto Navigli, Sapienza University of Rome, Italy