论文信息 - Correlation Voting Fusion Strategy for Part of Speech Tagging

Correlation Voting Fusion Strategy for Part of Speech Tagging

Having studied four corpus-based approaches to part of speech (POS) tagging, such as transform-based error driven, the decision tree, hidden Markov model and maximum entropy, we present in this paper a novel data fusion strategy in POS tagging - correlation voting. Theoretical analysis and contrastive experiments with other fusion strategies show that linguistic knowledge for POS tagging can be more completely described by applying data fusion, and better tagging result can be achieved. The correlative voting is proved to be more outstanding than other fusion methods with a decrease of 27.85% in average tagging error rate

[1] Eric Brill,et al. Unsupervised Learning of Disambiguation Rules for Part of Speech Tagging , 1995, VLC@ACL.

[2] Walter Daelemans,et al. Introduction to Special Issue on Machine Learning Approaches to Shallow Parsing , 2002, J. Mach. Learn. Res..

[3] Salvatore J. Stolfo,et al. A Comparative Evaluation of Voting and Meta-learning on Partitioned Data , 1995, ICML.

[4] Helmut Schmidt,et al. Probabilistic part-of-speech tagging using decision trees , 1994 .

[5] Thorsten Brants,et al. TnT – A Statistical Part-of-Speech Tagger , 2000, ANLP.

[6] Beatrice Santorini,et al. Building a Large Annotated Corpus of English: The Penn Treebank , 1993, CL.

[7] Adwait Ratnaparkhi,et al. A Maximum Entropy Model for Part-Of-Speech Tagging , 1996, EMNLP.

[8] Eric Brill,et al. A corpus-based approach to language learning , 1993 .