Paraphrasing, textual entailment, and semantic similarity above word level

TL;DR

This dissertation investigates the linguistic and computational aspects of semantic relations like paraphrasing, entailment, and similarity across various levels of language, proposing new insights and methodologies for understanding and identifying these relations.

Contribution

It introduces a comprehensive analysis of semantic relations at multiple linguistic levels and presents novel approaches to paraphrase identification and the joint study of related meaning relations.

Findings

Different methodologies for quantifying semantic similarity at word and phrase levels

Empirical insights into paraphrase typology and identification

A new integrated approach to studying textual entailment and semantic similarity

Abstract

This dissertation explores the linguistic and computational aspects of the meaning relations that can hold between two or more complex linguistic expressions (phrases, clauses, sentences, paragraphs). In particular, it focuses on Paraphrasing, Textual Entailment, Contradiction, and Semantic Similarity. In Part I: "Similarity at the Level of Words and Phrases", I study the Distributional Hypothesis (DH) and explore several different methodologies for quantifying semantic similarity at the levels of words and short phrases. In Part II: "Paraphrase Typology and Paraphrase Identification", I focus on the meaning relation of paraphrasing and the empirical task of automated Paraphrase Identification (PI). In Part III: "Paraphrasing, Textual Entailment, and Semantic Similarity", I present a novel direction in the research on textual meaning relations, resulting from joint research carried out on on paraphrasing, textual entailment, contradiction, and semantic similarity.