model,11-1-H01-1058,bq problem of combining several <term> language models ( LMs ) </term> . We find that simple <term>
tech,11-5-H01-1058,bq show the need for a <term> dynamic language model combination </term> to improve the <term> performance
model,7-1-H01-1070,bq practical approach employing <term> n-gram models </term> and <term> error-correction rules </term>
model,31-2-P01-1004,bq a range of <term> local segment contiguity models </term> ( in the form of <term> N-grams </term>
model,24-3-P01-1004,bq superior to any of the tested <term> word N-gram models </term> . Further , in their optimum <term>
other,14-1-P01-1070,bq on the construction of <term> statistical models </term> of <term> WH-questions </term> . These
model,1-2-P01-1070,bq of <term> WH-questions </term> . These <term> models </term> , which are built from <term> shallow
other,11-3-P01-1070,bq predictive performance </term> of our <term> models </term> , including the influence of various
model,3-2-N03-1001,bq combines <term> domain independent acoustic models </term> with off-the-shelf <term> classifiers
model,12-3-N03-1001,bq first used to train a <term> phone n-gram model </term> for a particular <term> domain </term>
model,26-3-N03-1001,bq of <term> recognition </term> with this <term> model </term> is then passed to a <term> phone-string
model,4-1-N03-1017,bq propose a new <term> phrase-based translation model </term> and <term> decoding algorithm </term>
model,21-1-N03-1017,bq previously proposed <term> phrase-based translation models </term> . Within our framework , we carry
model,18-2-N03-1017,bq better and explain why <term> phrase-based models </term> outperform <term> word-based models
model,21-2-N03-1017,bq models </term> outperform <term> word-based models </term> . Our empirical results , which hold
model,12-4-N03-1017,bq <term> high-accuracy word-level alignment models </term> does not have a strong impact on
model,7-1-N03-1018,bq probabilistic optical character recognition ( OCR ) model </term> that describes an end-to-end process
model,1-2-N03-1018,bq </term> of an <term> OCR system </term> . The <term> model </term> is designed for use in <term> error
model,6-3-N03-1018,bq We present an implementation of the <term> model </term> based on <term> finite-state models
tech,9-3-N03-1018,bq <term> model </term> based on <term> finite-state models </term> , demonstrate the <term> model </term>
hide detail