Paraphrase Generation from Latent-Variable PCFGs for Semantic Parsing

TL;DR

This paper introduces a novel latent-variable PCFG-based method for paraphrase generation to improve semantic parsing in open-domain question answering, significantly enhancing parser performance without needing aligned paraphrase data.

Contribution

It presents a new grammar model leveraging latent-variable PCFGs for paraphrase generation, addressing the lexicosyntactic gap in semantic parsing without requiring sentence-aligned corpora.

Findings

Improved semantic parser accuracy on WebQuestions dataset

Significant performance gains over baseline methods

Effective paraphrase sampling with latent-variable PCFGs

Abstract

One of the limitations of semantic parsing approaches to open-domain question answering is the lexicosyntactic gap between natural language questions and knowledge base entries -- there are many ways to ask a question, all with the same answer. In this paper we propose to bridge this gap by generating paraphrases of the input question with the goal that at least one of them will be correctly mapped to a knowledge-base query. We introduce a novel grammar model for paraphrase generation that does not require any sentence-aligned paraphrase corpus. Our key idea is to leverage the flexibility and scalability of latent-variable probabilistic context-free grammars to sample paraphrases. We do an extrinsic evaluation of our paraphrases by plugging them into a semantic parser for Freebase. Our evaluation experiments on the WebQuestions benchmark dataset show that the performance of the semantic parser significantly improves over strong baselines.