Publication at Faculty of Mathematics and Physics |
2016
Abstract
Recognizing textual entailment is typically considered as a binary decision task – whether a text T entails a hypothesis H. Thus, in case of a negative answer, it is not possible to express that H is “almost entailed” by T.
Partial textual entailment provides one possible approach to this issue. This paper presents an attempt to use word2vec model for recognizing partial (faceted) textual entailment.
The proposed approach does not rely on language dependent NLP tools and other linguistic resources, therefore it can be easily implemented in different language environments where word2vec models are available.