Reducing the Impact of Data Sparsity in Statistical Machine Translation
BibTeX
@MISC{Singla_reducingthe,
author = {Karan Singla and Kunal Sachdeva and Diksha Yadav and Srinivas Bangalore and Dipti Misra Sharma},
title = {Reducing the Impact of Data Sparsity in Statistical Machine Translation},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Morphologically rich languages generally require large amounts of parallel data to adequately estimate parameters in a statis-tical Machine Translation(SMT) system. However, it is time consuming and expen-sive to create large collections of parallel data. In this paper, we explore two strate-gies for circumventing sparsity caused by lack of large parallel corpora. First, we ex-plore the use of distributed representations in an Recurrent Neural Network based lan-guage model with different morphological features and second, we explore the use of lexical resources such as WordNet to over-come sparsity of content words. 1
Keyphrases
statistical machine translation data sparsity parallel data large collection lan-guage model rich language content word lexical resource different morphological feature over-come sparsity large amount statis-tical machine translation large parallel corpus distributed representation recurrent neural network
