Exploiting semantic role labeling, WordNet and Wikipedia for coreference resolution
HLT-NAACL '06 Proceedings of the main conference on Human Language Technology Conference of the North American Chapter of the Association of Computational Linguistics
| Hi-index | 0.00 |
Effectively collecting and extracting useful information from the relevant news reports are the keys to raise the response capabilities to handle emergencies. So an important subtask is how to resolve the coreference phenomenon. In this paper, we present a coreference resolution approach based on maximum entropy model in paroxysmal events news reports. The pronouns, nouns and noun phrases which refer to the same entity in a Chinese news report can be extracted by the approach. A group of semantic features are used to coreference resolution, which includes the semantic class features, the semantic role features based on the pronoun refining, the semantic-related features, the redirection features and context features based on Wikipedia. We compare their performance on the testing corpus. The training corpus contains 200,000 Chinese characters and the testing corpus contains 50,000. The experimental results show the improvement of coreference resolution after adding selected semantic knowledge.