E06-1006 ( Resnik et al. , 2001 ) used backoff translation lexicons for cross-language information
H01-1060 little surface morphology , we omit backoff translation in this case . ments : Bars above
H01-1033 translation lexicons , and show that backoff translation outperforms a technique based
N03-2026 translation architecture using backoff translation and Pirkola 's structured query
H01-1060 were then indexed with of 4-stage backoff translation with statistical stemming . the
W13-2207 form and tag , and the secondary backoff translation is from a " stem " described
H01-1060 effectiveness is illustrated in Figure 1 . Backoff translation improves translation coverage
H01-1033 Cross-Language Retrieval using Backoff Translation </title> Philip Resnik Douglas
H01-1060 processingcomponentsdescribed above - statistical stemming backoff translation - applied to the English - French
H01-1060 invested in development of four-stage backoff translation and statistical stemming . Further
P08-1115 lattices are similar in spirit to backoff translation models ( Yang and Kirchhoff ,
W12-3158 plan to extend our work to use backoff translations ( the target word that can not
H01-1060 to unstemmed ef - fectiveness . Backoff translation improves cross-languageinformation
H01-1033 Cross-Language Retrieval using Backoff Translation Philip Resnik , 1 ; ; 2 Douglas
N03-2026 version 3.1 p1 . The key idea in backoff translation is to first try to find consecutive
H01-1033 in our comparisonof merging and backoff translation strategies for combining resources
H01-1033 of balanced translation and a backoff translation strategy ( which produces a stronger
H01-1060 programmer time . 2.2 Four-Stage Backoff Translation . Bilingual term lists found
hide detail