logo
banner

News & Events

News & Events

CASIA Won 23rd PACLIC Best Paper Award
Dec 04, 2009Author:
PrintText Size A A

ZHANG Jiajun, a Ph.D. candidate in National Laboratory of Pattern Recognition (NLPR) and Prof. ZONG Chengqing, advisor of ZHANG Jiajun, have been awarded the best paper with their co-authored paper, 'A Framework for Effectively Integrating Hard and Soft Syntactic Rules into Phrase Based Translation' in the 23rd Pacific Asia Conference on Language, Information and Computation (PACLIC) on Oct. 5th.

PACLIC is an influential international conference of a long history in the field of Nature Language Processing. It is ranked at the top 7% in international conferences about Artificial Intelligence and Machine Learning, according to the Computer Science Conference Ranking Website. The 23rd of the annual event was held successfully in Hong Kong in Oct. 3rd to 5th this year. This PACLIC has received 145 submissions, 58 of which were accepted and 2 of which were awarded the best paper.

Current statistical machine translation methods suffer from propagating the pre-reordering errors to the later translation step. ZHANG Jiajun and ZONG Chengqing proposed a novel framework to integrate hard and soft syntactic rules into phrase-based translation more effectively aiming at this probelm. The basic idea is: For a source sentence to be translated, hard or soft syntactic rules are first acquired from the source parse tree prior to translation, and then instead of reordering the source sentence directly, the rules are used as a strong feature integrated into our elaborately designed model to help phrase reordering in the decoding stage. The experiments on NIST Chinese-to-English translation show that this approach significantly improves translation quality. Program committee of the 23d PACLIC highly commented on this paper. It can achieve remarkable translation performance without additional time consuming compared with traditional methods. It is suitable for translation between most Asian languages and European languages.