摘要: A maximum entropy classifier can be used to extract sentences from documents. Experiments using technical documents show that such a tends treat features in categorical manner. This results performance is worse than when extracting naive Bayes classifier. Addition of an optimised prior the improves over and above (even also extended with similar prior). Further experiments that, should we have at our disposal extremely informative features, then able yield excellent results. Naive Bayes, contrast, cannot exploit these so fundamentally limits sentence extraction performance.
aclweb.org
acm.org参考文章(14)
John Lafferty, Kamal Nigam, Andrew McCallum, Using Maximum Entropy for Text Classification ,(1999)
Kamal Nigam, Andrew McCallum, A comparison of event models for naive bayes text classification national conference on artificial intelligence. pp. 41- 48 ,(1998)
Stanley F Chen, Ronald Rosenfeld, A Gaussian Prior for Smoothing Maximum Entropy Models Defense Technical Information Center. ,(1999) , 10.21236/ADA360974
Daniel Marcu, The automatic construction of large-scale corpora for summarization research international acm sigir conference on research and development in information retrieval. pp. 137- 144 ,(1999) , 10.1145/312624.312668
Branimir K. Boguraev, Mary S. Neff, The effects of analysing cohesion on document summarisation international conference on computational linguistics. pp. 76- 82 ,(2000) , 10.3115/990820.990832
Jade Goldstein, Mark Kantrowitz, Vibhu Mittal, Jaime Carbonell, Summarizing text documents: sentence selection and evaluation metrics international acm sigir conference on research and development in information retrieval. pp. 121- 128 ,(1999) , 10.1145/312624.312665
Avrim Blum, Tom Mitchell, None, Combining labeled and unlabeled data with co-training conference on learning theory. pp. 92- 100 ,(1998) , 10.1145/279943.279962
Vincent J. Della Pietra, Adam L. Berger, Stephen A. Della Pietra, A maximum entropy approach to natural language processing Computational Linguistics. ,vol. 22, pp. 39- 71 ,(1996) , 10.5555/234285.234289
Julian Kupiec, Jan Pedersen, Francine Chen, A trainable document summarizer Proceedings of the 18th annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '95. pp. 68- 73 ,(1995) , 10.1145/215206.215333
Pedro Domingos, Michael Pazzani, On the Optimality of the Simple Bayesian Classifier under Zero-One Loss Machine Learning. ,vol. 29, pp. 103- 130 ,(1997) , 10.1023/A:1007413511361