| lr,7-3-H94-1014,ak | </term> were constructed using a 5K <term> | vocabulary | </term> and trained using a 76 million word | #27134 The models were constructed using a 5Kvocabulary and trained using a 76 million word Wall Street Journal text corpus. | |
| other,17-1-H94-1014,ak | long distance constraints </term> in a <term> | sentence | </term> or <term> paragraph </term> . The <term> | #27113 This paper introduces a simple mixture language model that attempts to capture long distance constraints in asentence or paragraph. | |
| model,25-4-H94-1014,ak | models </term> as compared to using a <term> | trigram model | </term> . This paper describes a method of | #27173 Using the BU recognition system, experiments show a 7% improvement in recognition accuracy with the mixture trigram models as compared to using atrigram model. | |
| other,4-2-H94-1014,ak | </term> . The <term> model </term> is an <term> | m-component mixture | </term> of <term> trigram models </term> . The | #27121 The model is anm-component mixture of trigram models. | |
| other,12-1-H94-1014,ak | model </term> that attempts to capture <term> | long distance constraints | </term> in a <term> sentence </term> or <term> | #27108 This paper introduces a simple mixture language model that attempts to capturelong distance constraints in a sentence or paragraph. | |
| measure(ment),13-4-H94-1014,ak | experiments show a 7 % improvement in <term> | recognition accuracy | </term> with the <term> mixture trigram models | #27161 Using the BU recognition system, experiments show a 7% improvement inrecognition accuracy with the mixture trigram models as compared to using a trigram model. | |
| model,7-2-H94-1014,ak | <term> m-component mixture </term> of <term> | trigram models | </term> . The <term> models </term> were constructed | #27124 The model is an m-component mixture oftrigram models. | |
| other,19-1-H94-1014,ak | </term> in a <term> sentence </term> or <term> | paragraph | </term> . The <term> model </term> is an <term> | #27115 This paper introduces a simple mixture language model that attempts to capture long distance constraints in a sentence orparagraph. | |
| model,5-1-H94-1014,ak | </term> . This paper introduces a simple <term> | mixture language model | </term> that attempts to capture <term> long | #27101 This paper introduces a simplemixture language model that attempts to capture long distance constraints in a sentence or paragraph. | |
| model,1-3-H94-1014,ak | of <term> trigram models </term> . The <term> | models | </term> were constructed using a 5K <term> | #27128 Themodels were constructed using a 5K vocabulary and trained using a 76 million word Wall Street Journal text corpus. | |
| model,1-2-H94-1014,ak | </term> or <term> paragraph </term> . The <term> | model | </term> is an <term> m-component mixture </term> | #27118 Themodel is an m-component mixture of trigram models. | |
| tool,2-4-H94-1014,ak | Journal text corpus </term> . Using the <term> | BU recognition system | </term> , experiments show a 7 % improvement | #27150 Using theBU recognition system, experiments show a 7% improvement in recognition accuracy with the mixture trigram models as compared to using a trigram model. | |
| model,17-4-H94-1014,ak | recognition accuracy </term> with the <term> | mixture trigram models | </term> as compared to using a <term> trigram | #27165 Using the BU recognition system, experiments show a 7% improvement in recognition accuracy with themixture trigram models as compared to using a trigram model. | |
| lr-prod,15-3-H94-1014,ak | and trained using a 76 million word <term> | Wall Street Journal text corpus | </term> . Using the <term> BU recognition system | #27142 The models were constructed using a 5K vocabulary and trained using a 76 million wordWall Street Journal text corpus. |
