Using Parsed Corpora for Structural Disambiguation in the TRAINS Domain

This paper describes a prototype disambiguation module KANKEI which uses two corpora of the TRAINS project In ambiguous verb phrases of form V NP PP or V NP adverb s the two corpora have very di erent PP and adverb attachment patterns in the rst the correct attachment is to the VP of the time while in the second the correct attachment is to the NP of the time KANKEI uses various n gram patterns of the phrase heads around these ambiguities and assigns parse trees with these ambiguities a score based on a linear combination of the frequencies with which these patterns appear with NP and VP attachments in the TRAINS corpora Unlike previous statistical disambiguation systems this technique thus combines evidence from bigrams trigrams and the gram around an ambiguous attachment In the current experiments equal weights are used for simplicity but results are still good on the TRAINS corpora and accuracy Despite the large statistical di erences in attachment preferences in the two corpora training on the rst corpus and testing on the second gives an accuracy of These results suggest that our technique captures attachment patterns that are useful across corpora This work was supported in part by National Science Foundation grant IRI James Allen s helpful suggestions are gratefully acknowledged