January 1, 1992

Cooccurrence smoothing for stochastic language modeling

Key Points

Key points are not available for this paper at this time.

Abstract

Training corpora for stochastic language models are virtually always too small for maximum-likelihood estimation, so smoothing the models is of great importance. The authors derive the cooccurrence smoothing technique for stochastic language modeling and give experimental evidence for its validity. Using word-bigram language models, cooccurrence smoothing improved the test-set perplexity by 14% on a German 100000-word text corpus and by 10% on an English 1-million word corpus.>

Perguntar à IA

Bookmark

Cite This Study

Essen et al. (Wed,) studied this question.

synapsesocial.com/papers/6a183c683aabde875b1375b8 https://doi.org/https://doi.org/10.1109/icassp.1992.225947

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

1On smoothing techniques for bigram-based natural language modelling1991 · 73 citations
2Isolated word recognition using hidden Markov models2005 · 35 citations
3Robust smoothing methods for discrete hidden Markov models2003 · 32 citations
4A Maximum Likelihood Approach to Continuous Speech Recognition1983 · 1,221 citations
5Estimation of probabilities from sparse data for the language model component of a speech recognizer1987 · 1,653 citations

Perguntar à IA

Bookmark