Correcting N-gram probabilities by page view information

A non-transitory computer readable storage medium comprising a computer readable program for a method for calculating N-gram probabilities in a language model, wherein the computer readable program when executed on a computer causes the computer to perform the following method steps:

• sorting a plurality of pages or a plurality of documents based on at least one of view counts and rankings to obtain sorted pages or sorted documents;
• classifying the sorted pages or the sorted documents into M subsets;
• counting N-grams in each of the M-subsets;
• estimating respective N-gram probabilities for each of the N-grams in each of the M-subsets using a maximum likelihood method;
• interpolating the estimated respective N-gram probabilities to minimize a cross entropy there between;
• and
• decoding an utterance, by a speech recognition system, based on the N-gram probabilities in the language model.

A system for calculating N-gram probabilities in a language model, comprising:

• an N-gram counter for counting N-grams in each page of a plurality of pages or in each document of a plurality of documents to obtain respective N-gram counts therefor;
• a weighted N-gram counter for applying weights to the respective N-gram counts based on at least one of view counts and rankings to obtain weighted respective N-gram counts, the view counts and the rankings determined with respect to the plurality of pages or the plurality of documents;
• an N-gram merger for merging the weighted respective N-gram counts to obtain merged weighted respective N-gram counts for the plurality of pages or the plurality of documents;
• a probability calculator for calculating a respective probability for each of the N-grams based on the merged weighted respective N-gram counts;
• and
• a speech recognition system for decoding an utterance based on the N-gram probabilities in the language model.