Paper: Semi-Supervised Learning for Semantic Parsing using Support Vector Machines

ACL ID N07-2021
Title Semi-Supervised Learning for Semantic Parsing using Support Vector Machines
Venue Human Language Technologies
Session Short Paper
Year 2007
Authors

We present a method for utilizing unan- notated sentences to improve a semantic parser which maps natural language (NL) sentences into their formal meaning rep- resentations (MRs). Given NL sentences annotated with their MRs, the initial su- pervised semantic parser learns the map- ping by training Support Vector Machine (SVM) classifiers for every production in the MR grammar. Our new method ap- plies the learned semantic parser to the unannotated sentences and collects unla- beled examples which are then used to retrain the classifiers using a variant of transductive SVMs. Experimental results show the improvements obtained over the purely supervised parser, particularly when the annotated training set is small.