🤖 AI Summary
Existing semantic parsers struggle to effectively leverage the rich structured information in knowledge bases (KBs), limiting both accuracy and efficiency in logical form generation. To address this, we propose a KB-enhanced grammar modeling framework that explicitly incorporates candidate entities and relations into the grammatical constraints of a seq2seq pretrained language model. Our approach introduces novel subtype inference and union-type rules, and designs a masked caching mechanism to accelerate constrained decoding. It integrates type-aware decoding, candidate-driven grammar modeling, and efficient inference strategies. Evaluated on KQA Pro and Overnight benchmarks under both strong and weak supervision, our method achieves significant improvements in logical form accuracy while substantially accelerating decoding—demonstrating its effectiveness, generalizability, and practical utility.
📝 Abstract
Semantic parsers convert natural language to logical forms, which can be evaluated on knowledge bases (KBs) to produce denotations. Recent semantic parsers have been developed with sequence-to-sequence (seq2seq) pre-trained language models (PLMs) or large language models, where the models treat logical forms as sequences of tokens. For syntactic and semantic validity, the semantic parsers use grammars that enable constrained decoding. However, the grammars lack the ability to utilize large information of KBs, although logical forms contain representations of KB elements, such as entities or relations. In this work, we propose a grammar augmented with candidate expressions for semantic parsing on a large KB with a seq2seq PLM. The grammar defines actions as production rules, and our semantic parser predicts actions during inference under the constraints by types and candidate expressions. We apply the grammar to knowledge base question answering, where the constraints by candidate expressions assist a semantic parser to generate valid KB elements. We also introduce two special rules, sub-type inference and union types, and a mask caching algorithm. In particular, sub-type inference and the mask caching algorithm greatly increase the decoding speed of our semantic parser. We experimented on two benchmarks, KQA Pro and Overnight, where the constraints by candidate expressions increased the accuracy of our semantic parser, whether it was trained with strong supervision or weak supervision. In addition, our semantic parser had a fast decoding speed in the experiments.