2015/091 - Semantic Relatedness and Textual Entailment via Corpus Patterns
We present a system for resolving both semantic relatedness (SR) and textual entailment (TE) tasks. There are two major contributions the method proposed here brings to the field:(1) it shows that there is a correlation between the SR scores and TE judgments which can be used to improve the accuracy of bothof these tasks and (2) it shows that we can handle the structural information via patterns extracted from corpora and that this approach brings a substantial improvement to distributional systems.
The system attains a new state of the art for TE and reaches a correlation score within 1.5% percent to the SR state of the art.
Citation: CLIN, Amsterdam, December 18th, 2015.