Inferring Logical Forms From Denotations
This addresses the challenge of improving semantic parsing accuracy for question-answering systems, representing an incremental advance with specific gains in coverage and spurious form reduction.
The paper tackles the problem of learning semantic parsers from denotations by efficiently handling a more expressive class of logical forms and filtering out spurious ones, increasing answerable question coverage from 53.5% to 76% and ruling out 92.1% of spurious logical forms on the WikiTableQuestions dataset.
A core problem in learning semantic parsers from denotations is picking out consistent logical forms--those that yield the correct denotation--from a combinatorially large space. To control the search space, previous work relied on restricted set of rules, which limits expressivity. In this paper, we consider a much more expressive class of logical forms, and show how to use dynamic programming to efficiently represent the complete set of consistent logical forms. Expressivity also introduces many more spurious logical forms which are consistent with the correct denotation but do not represent the meaning of the utterance. To address this, we generate fictitious worlds and use crowdsourced denotations on these worlds to filter out spurious logical forms. On the WikiTableQuestions dataset, we increase the coverage of answerable questions from 53.5% to 76%, and the additional crowdsourced supervision lets us rule out 92.1% of spurious logical forms.