A Supervised Semantic Parsing with Lexicon Extension and Syntactic Constraint

Existing semantic parsing research has steadily improved accuracy on a few domains and their corresponding meaning representations. In this paper, we present a novel supervised semantic parsing algorithm, which includes the lexicon extension and the syntactic supervision. This algorithm adopts a large-scale knowledge base from the open-domain Freebase to construct efficient, rich Combinatory Categorial Grammar (CCG) lexicon in order to supplement the inadequacy of its manually-annotated training dataset in the small closed-domain while allows for the syntactic supervision from the dependency-parsed sentences to penalize the ungrammatical semantic parses. Evaluations on both benchmark closed-domain datasets demonstrate that this approach learns highly accurate parser, whose parsing performance benefits greatly from the open-domain CCG lexicon and syntactic constraint.