Abstract

We present a system for deciding whether a given sentence can be inferred from text. Each sentence is represented as a directed graph (extracted from a dependency parser) in which the nodes represent words or phrases, and the links represent syntactic and semantic relationships. We develop a learned graph matching model to approximate entailment by the amount of the sentence’s semantic content which is contained in the text. We present results on the Recognizing Textual Entailment dataset (Dagan et al., 2005), and show that our approach outperforms Bag-Of-Words and TF-IDF models. In addition, we explore common sources of errors in our approach and how to remedy them. 1

Citations

1996	WordNet: An Electronic Lexical Database - Fellbaum, editor - 1998
992	BLEU: A Method for Automatic Evaluation of Machine Translation - Papineni, Roukos, et al. - 2002
780	Head-Driven Statistical Models for Natural Language Parsing - Collins - 1999
526	Using Information Content to Evaluate Semantic Similarity in a Taxonomy - Resnik
422	Accurate unlexicalized parsing - Klein, Manning - 2003
110	Dirt - discovery of inference rules from text - Dekang, Pantel - 2001
101	Verb ocean: Mining the web for fine-grained semantic verb relations - Chklovski, Pantel - 2004
65	S.: Cogex: A logic prover for question answering - Moldovan, Clark, et al. - 2003
7	textual inference via learning and abductive reasoning - Robust
1	Robust textual inference using diverse knowledge sources - 2005a