A maximum entropy approach to natural language processing
Computational Linguistics
Maximum entropy models for natural language ambiguity resolution
Maximum entropy models for natural language ambiguity resolution
Word-sense disambiguation using statistical models of Roget's categories trained on large corpora
COLING '92 Proceedings of the 14th conference on Computational linguistics - Volume 2
Word sense disambiguation using Conceptual Density
COLING '96 Proceedings of the 16th conference on Computational linguistics - Volume 1
HLT '93 Proceedings of the workshop on Human Language Technology
HLT '93 Proceedings of the workshop on Human Language Technology
Conditioning algorithms for exact and approximate inference in causal networks
UAI'95 Proceedings of the Eleventh conference on Uncertainty in artificial intelligence
The infocious web search engine: improving web searching through linguistic analysis
WWW '05 Special interest tracks and posters of the 14th international conference on World Wide Web
Semantic passage segmentation based on sentence topics for question answering
Information Sciences: an International Journal
Word Clustering for Collocation-Based Word Sense Disambiguation
CICLing '07 Proceedings of the 8th International Conference on Computational Linguistics and Intelligent Text Processing
PKU: combining supervised classifiers with features selection
SemEval '07 Proceedings of the 4th International Workshop on Semantic Evaluations
Hi-index | 0.00 |
A maximum entropy-based word sense disambiguation system is presented, consisting of individual word experts that are trained on both labeled and partially labeled corpora. The classification probabilities from the individual word experts are integrated using a new search algorithm, which balances time complexity and accuracy. The model is evaluated using established procedures on the English-all-words task from the SENSEVAL-2 workshop, a large test set consisting of words from all word groups to be disambiguated. Lastly, an ongoing project that integrates POS tagging, parsing, and sense disambiguation in one system is presented. Once in place, it will be boot-strapped with existing partially labeled corpora, to process and then train from them. The goal is to show that with each successive iteration, the accuracy of all three processes, POS tagging, parsing, and WSD, will improve as the system learns from more accurate, self-generated training data.