期刊名称：Indian Journal of Computer Science and Engineering
印刷版ISSN：2231-3850
电子版ISSN：0976-5166
出版年度：2011
卷号：2
期号：1
页码：94-100
出版社：Engg Journals Publications
摘要：In this paper, a recently proposed word alignment algorithm is simplified for easy understanding and tested for an Indian language. The word alignment problem is viewed as a simple assignment problem and is formulated as an Integer Linear Programming problem. The new objective function defined is tested for obtaining optimal alignment for English-Tamil translation pair. This alignment is necessary for creating the probabilistic bilingual dictionary and is also required for automatic machine translation. We have used this objective function to align words in 25 sentences of English-Tamil parallel corpora. The formulation is solved using the open source LP-Solver. Result obtained indicates that the methodology is applicable for all Indian languages. The work implemented is useful for pedagogical purposes, as it is a standard problem in computational linguistics. Accuracy of modern statistical machine translation depends on good word alignment. The document of the formulated model is available on request.
关键词：Word alignment; assignment problem; integer linear programming; alignment variables and dictionary variables.
