Abstract
This paper presents an approach which improves the performance of word alignment for English-Hindi language pair. Longer sentences in the corpus create severe problems like the high computational requirements and poor quality of resulting word alignment. Here, we present a method to solve these problems by breaking the longer sentence pairs into shorter ones. Our approach first breaks the source and target sentences into clauses and then treats the resulting clause pairs as sentence pairs to train word alignment model. We also report preliminary work on automatically identifying clause boundaries which are appropriate for improvement of word alignment. This paper demonstrates the increase of precision, recall and F-measure by approximately 11%, 7%, 10% respectively and reduction in Alignment Error Rate (AER) by approximately 10% in the performance of IBM Model 1 for word alignment. These results are obtained by training on 270 sentence pair and testing on 30 sentence pairs. Experiments of this paper are based on TDIL corpus.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
