Paper: How Well can We Learn Interpretable Entity Types from Text?

ACL ID P14-2079
Title How Well can We Learn Interpretable Entity Types from Text?
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2014
Authors

Many NLP applications rely on type sys- tems to represent higher-level classes. Domain-specific ones are more informa- tive, but have to be manually tailored to each task and domain, making them in- flexible and expensive. We investigate a largely unsupervised approach to learning interpretable, domain-specific entity types from unlabeled text. It assumes that any common noun in a domain can function as potential entity type, and uses those nouns as hidden variables in a HMM. To con- strain training, it extracts co-occurrence dictionaries of entities and common nouns from the data. We evaluate the learned types by measuring their prediction ac- curacy for verb arguments in several do- mains. The results suggest that it is pos- sible to learn domain-specific entity types from unlabeled data...