Paper: MT Error Detection for Cross-Lingual Question Answering

ACL ID C10-2109
Title MT Error Detection for Cross-Lingual Question Answering
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2010
Authors

We present a novel algorithm for de- tecting errors in MT, specifically focus- ing on content words that are deleted during MT. We evaluate it in the con- text of cross-lingual question answering (CLQA), where we try to correct the detected errors by using a better (but slower) MT system to retranslate a lim- ited number of sentences at query time. Using a query-dependent ranking heuris- tic enabled the system to direct scarce MT resources towards retranslating the sentences that were most likely to ben- efit CLQA. The error detection algo- rithm identified spuriously deleted con- tent words with high precision. How- ever, retranslation was not an effective approach for correcting them, which in- dicates the need for a more targeted ap- proach to error correction in the future.