A study on textual entailment

Abstract

This paper studies the role of dependency information for the task of textual entailment. Both the Text and Hypothesis of an entailment pair are mapped into sets of dependencies and a score is computed that measures the similarity of the two sets. Based on the score an entailment decision is made. Two experiments are conducted to measure the impact of dependencies on the entailment task. In one experiment we compare the dependency-based approach with a baseline approach on a standard data set. In a second experiment, we measure the performance on a subset of the standard data set. The subset is so selected to minimize the effect of other factors, such as word-level information, in the performance measurement process. A brief discussion compares the dependency-based approach to other, similar approaches.

AutoTutor and affective autotutor

“…We are working on systems that analyze language and discourse at deeper levels. Researchers can move beyond LSA and regular expressions and into more structure-sensitive processing and semantic decomposition [Rus and Graesser 2007;Rus et al 2008]. Some versions of the dialog manager module have lists of production rules and finite-state grammars, but also move to new levels into the realm of recursive, complex planning and multiple-goal agendas.…”

Identification of Sentence-to-Sentence Relations Using a Textual Entailer

“…Despite such successes, the need remains for new measures of textual assessment to augment existing measures and thereby better assess textual comparisons. We assess in this article a variety of more established textual relatedness assessment metrics (e.g., LSA and Content-Overlap), and compare them to newer approaches such as the Entailer (Rus et al 2008) and MED (McCarthy et al 2007(McCarthy et al , 2009. Each of these measures provides a unique approach to assessing the relatedness between text fragments.…”

“…In the preprocessing stage, the system separates the punctuation from words (tokenization), maps morphological variations of words to their base or root form (lemmatization), assigns part-of-speech labels to each word (tagging), and assesses the inter-relationship of major phrases within the texts (parsing). Additional preprocessing operations are also performed, such as detecting collocations of common nouns (for more details see Rus et al 2008). For instance, joint venture is identified as a collocation and thus a single node in the graph is generated for it instead of two, i.e., one for joint and one for venture.…”

“…The paraphrase identification approach is based on a previously proposed approach to entailment that offered competitive results as compared to other approaches on a Recognizing Textual Entailment dataset (RTE; Dagan et al 2004Dagan et al -2005. The entailment approach (Rus et al 2008) uses lexical, syntactic, synonymy and antonymy (for negation handling) information. The synonymy and antonymy information is extracted from a thesaurus, i.e., WordNet (Miller 1995).…”