Predicting Correlations Between Lexical Alignments and Semantic Inferences

While there is a strong intuition that word alignments (e.g. synonymy, hyperonymy) play a relevant role in recognizing text-to-text semantic inferences (e.g. textual entailment, semantic similarity), this intuition is often not reflected in the system performances and there is a general need of a deeper comprehension of the role of lexical resources. This paper provides an empirical analysis of the dependencies between data-sets, lexical resources and algorithms that are commonly used in text-to-text inference tasks. We define a resource impact index , based on lexical alignments between pairs of texts, and show that such index is significantly correlated with the performance of different textual entailment algorithms. The result is an operational, algorithm-independent, procedure for predicting the performance of a class of available RTE algorithms.