Logic-Consistency Text Generation from Semantic Parses
This addresses the challenge of generating accurate textual descriptions from formal representations like logic forms and SQL queries, which is important for applications in natural language generation and human-computer interaction, though it is an incremental improvement on existing methods.
The paper tackles the problem of generating logically consistent text from semantic parses by proposing SNOWBALL, an iterative training framework that improves logic consistency by 5-10% on BLEC scores, and BLEC, a novel automatic metric that correlates better with human evaluation than existing metrics like BLEU and ROUGE.
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.