New approach for collecting high quality parallel corpora from multilingual websites
Cong Phap Huynh · 2011
In this paper, we present a new approach for extracting the high quality (HQ) parallel corpora from multilingual resources. The original of our research compared to the previous works is the approach for gaining HQ data using for the Machine Translation domain. Almost previous approaches allowed to quickly acquire raw corpora, but not allow to gain HQ data. Our approach is a semi-automatic process including in a serial of steps that can automatically detect and download good multilingual Websites and parallel web pages to construct parallel corpora whose quality is well validated, revised, and enhanced collaboratively.