Context dependent class language model based on word co-occurrence matrix in LSA framework for speech recognition Export

@inproceedings{citeulike:5138592, abstract = {We address the issue of data sparseness problem in language model (LM). Using class LM is one way to avoid this problem. In class LM, infrequent words are supported by more frequent words in the same class. This paper investigates a class LM based on LSA. A word-document matrix is usually used to represent a corpus in LSA framework. However, this matrix ignores word order in the sentence. We propose several word co-occurrence matrices that keep word order. Together with these matrices, we define a context dependent class (CDC) LM which distinguishes classes according to their context in the sentences. Experiments on Wall Street Journal (WSJ) corpus show that the word co-occurrence matrix works better than word-document matrix. Furthermore, the CDC achieves better perplexity than the traditional class LM based on LSA.}, address = {Stevens Point, Wisconsin, USA}, author = {Naptali, Welly and Tsuchiya, Masatoshi and Nakagawa, Seiichi}, booktitle = {ACS'08: Proceedings of the 8th conference on Applied computer scince}, citeulike-article-id = {5138592}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1504085}, citeulike-linkout-1 = {http://www.slp.ics.tut.ac.jp/shiryou/extra_2008/2008-IC-09.pdf}, isbn = {978-960-474-028-4}, keywords = {class, lm, lsa}, location = {Venice, Italy}, pages = {275--280}, posted-at = {2009-07-13 18:41:05}, priority = {0}, publisher = {World Scientific and Engineering Academy and Society (WSEAS)}, title = {Context dependent class language model based on word co-occurrence matrix in LSA framework for speech recognition}, url = {http://portal.acm.org/citation.cfm?id=1504085}, year = {2008} }

TY - CONF ID - citeulike:5138592 L3 - citeulike-article-id:5138592 N2 - We address the issue of data sparseness problem in language model (LM). Using class LM is one way to avoid this problem. In class LM, infrequent words are supported by more frequent words in the same class. This paper investigates a class LM based on LSA. A word-document matrix is usually used to represent a corpus in LSA framework. However, this matrix ignores word order in the sentence. We propose several word co-occurrence matrices that keep word order. Together with these matrices, we define a context dependent class (CDC) LM which distinguishes classes according to their context in the sentences. Experiments on Wall Street Journal (WSJ) corpus show that the word co-occurrence matrix works better than word-document matrix. Furthermore, the CDC achieves better perplexity than the traditional class LM based on LSA. CY - Venice, Italy SN - 978-960-474-028-4 AD - Stevens Point, Wisconsin, USA EP - 280 TI - Context dependent class language model based on word co-occurrence matrix in LSA framework for speech recognition PB - World Scientific and Engineering Academy and Society (WSEAS) T2 - ACS'08: Proceedings of the 8th conference on Applied computer scince SP - 275 KW - class KW - lm KW - lsa AU - Naptali, Welly AU - Tsuchiya, Masatoshi AU - Nakagawa, Seiichi PY - 2008/// UR - http://portal.acm.org/citation.cfm?id=1504085 ER -