Logic-Consistency Text Generation from Semantic Parses

TL;DR

This paper introduces SNOWBALL, a framework for generating logically consistent text from semantic parses, and proposes BLEC, an automatic metric for evaluating logic consistency, demonstrating improved performance on benchmark datasets.

Contribution

The paper presents a novel iterative training framework and a new automatic metric for logic consistency in text generation from semantic parses.

Findings

SNOWBALL improves logic consistency on benchmark datasets.

BLEC correlates better with human judgment than existing metrics.

The approach enhances both BLEC scores and human evaluation results.

Abstract

Text generation from semantic parses is to generate textual descriptions for formal representation inputs such as logic forms and SQL queries. This is challenging due to two reasons: (1) the complex and intensive inner logic with the data scarcity constraint, (2) the lack of automatic evaluation metrics for logic consistency. To address these two challenges, this paper first proposes SNOWBALL, a framework for logic consistent text generation from semantic parses that employs an iterative training procedure by recursively augmenting the training set with quality control. Second, we propose a novel automatic metric, BLEC, for evaluating the logical consistency between the semantic parses and generated texts. The experimental results on two benchmark datasets, Logic2Text and Spider, demonstrate the SNOWBALL framework enhances the logic consistency on both BLEC and human evaluation. Furthermore, our statistical analysis reveals that BLEC is more logically consistent with human evaluation than general-purpose automatic metrics including BLEU, ROUGE and, BLEURT. Our data and code are available at https://github.com/Ciaranshu/relogic.