P15-2103 discuss the intuitions of proposed language model smoothing . Generally , given a non-smoothed
D09-1078 conducted our experiments on seven language model smoothing methods . Five of these are well-known
P15-2103 performance of all methods of language model smoothing on the Twitter dataset - s .
D12-1107 storage size . 3.2 Less Memory Many language model smoothing strategies , including modified
P15-2103 incorporates social factors in language model smoothing . There is a study in ( Lin et
N04-1039 smoothing , the best performing language model smoothing technique . This justification
D09-1078 significance-based N-gram selection for seven language model smoothing meth - ods . For the best three
P15-2103 evaluate the effect of our proposed language model smoothing model using datasets from Twit
P15-2103 factor is important and unique for language model smoothing on social net - works . It should
P15-2103 framework with regularization for language model smoothing on social networks , using both
P15-2103 works . Then we introduce the language model smoothing with social regularization and
P15-2103 intuitions , document structure based language model smoothing is another direction to investigate
P15-2103 work - s , we have proposed a language model smoothing framework which incorporates
P15-2103 2005 ) , and we could form the language model smoothing under this optimization frame
W04-3242 seen in the training data . 2.1 Language Model Smoothing An - gram model when is called
D09-1078 perplexity when applied to a number of language model smoothing methods , including the widely-used
P15-2103 ( wz ) 5 Conclusions impact on language model smoothing . We make a further comparison
P06-1129 proposed to perform tasks such as language model smoothing and word clustering , but to
P06-1129 acquisition ( Dekang Lin , 1998 ) and language model smoothing ( Essen and Steinbiss , 1992
P15-2103 • We have proposed a balanced language model smoothing framework with optimization ,
hide detail