Natural Answer Generation: From Factoid Answer to Full-length Answer using Grammar Correction

TL;DR

This paper introduces a novel system for generating full-length answers from factoid answers in question answering, utilizing parse trees and grammar correction to improve fluency and accuracy, outperforming state-of-the-art methods.

Contribution

The proposed system combines parse tree analysis with a transformer-based grammar correction model to produce more fluent and accurate full-length answers in a domain-independent manner.

Findings

Outperforms SOTA in ROUGE-1 scores on NewsQA and SqUAD datasets.

Reduces inference time by 85% compared to SOTA.

Effective on both factoid and existential questions.

Abstract

Question Answering systems these days typically use template-based language generation. Though adequate for a domain-specific task, these systems are too restrictive and predefined for domain-independent systems. This paper proposes a system that outputs a full-length answer given a question and the extracted factoid answer (short spans such as named entities) as the input. Our system uses constituency and dependency parse trees of questions. A transformer-based Grammar Error Correction model GECToR (2020), is used as a post-processing step for better fluency. We compare our system with (i) Modified Pointer Generator (SOTA) and (ii) Fine-tuned DialoGPT for factoid questions. We also test our approach on existential (yes-no) questions with better results. Our model generates accurate and fluent answers than the state-of-the-art (SOTA) approaches. The evaluation is done on NewsQA and SqUAD datasets with an increment of 0.4 and 0.9 percentage points in ROUGE-1 score respectively. Also the inference time is reduced by 85\% as compared to the SOTA. The improved datasets used for our evaluation will be released as part of the research contribution.