Abstract

In this paper, we present empirical results on the challenge of learning to read. That is, given a handful of examples of the concepts and relations in an ontology and a large corpus, the system should learn to map from text to the concepts/relations of the ontology. In this paper, we report contrastive experiments on the recall, precision, and F-measure (F) of the mapping in the following conditions:(1) employing word-based patterns, employing semantic structure, and combining the two; and (2) fully automatic learning versus allowing minimal questions of a human informant.

Date

2010

Authors

Marjorie Freedman, Edward Loper, Elizabeth Boschee, Ralph Weischedel

Conference

Proceedings of the NAACL HLT 2010 First International Workshop on Formalisms and Methodology for Learning by Reading

Pages

61-69