Textual entailment through extended lexical overlap and lexico-semantic matching

Abstract

This paper presents two systems for textual entailment, both employing decision trees as a supervised learning algorithm. The first one is based primarily on the concept of lexical overlap, considering a bag of words similarity overlap measure to form a mapping of terms in the hypothesis to the source text. The second system is a lexicosemantic matching between the text and the hypothesis that attempts an alignment between chunks in the hypothesis and chunks in the text, and a representation of the text and hypothesis as two dependency graphs. Their performances are compared and their positive and negative aspects are analyzed.

Textual Entailment Using Lexical And Syntactic Similarity

“…They have tested different system settings for calculating the importance of the words of the hypothesis and investigated the possibility of combining them with machine learning algorithm. The system described in [19] consists of a bag of words similarity overlap measure, derived from a combination of WordNet lexical chains to form a mapping of terms in the hypothesis to the source text. These items were entered into a decision tree to determine the overall entailment relation.…”

Augmenting WordNet for deep understanding of text

“…Many existing RTE systems, e.g., (Adams et al, 2007;Chambers et al, 2007) largely work by statistically scoring the match between T and H, but this to an extent sidesteps "deep" language understanding, namely building a coherent, internal representation of the overall scenario the input text was intended to convey. RTE is one way of measuring success in this endeavor, but it is also possible to do moderately well in RTE without the system even attempting to "understand" the scenario the text is describing.…”

DLSITE-1: Lexical Analysis for Solving Textual Entailment Recognition

“…For example, in [1] the author uses WordNet in order to obtain the lexical relation between two tokens as well as a negation detector. The approach in [7] combines lexico-semantic information obtained by a large collection of paraphrases and NLP applications (e.g.…”

“…Otherwise, no increment is produced. Finally, this weight is normalized dividing it by the length of the hypothesis, calculated as the number of words, as shown in Equation 1.…”

“…The systems that participated in the challenges used different strategies that combined a wide variety of NLP techniques in order to detect textual entailment. For instance, it is clearly stated that the use of ngrams and subsequence overlap [12,5], syntactic matching [8], logical inference [4,13] and Machine Learning (ML) classification [4,1] is quite appropriate for identifying entailment inferences.…”