| 摘要 |
A method and apparatus for example-based spoken language translation with examples having grades of specificity are provided. A speech input is received comprising source language expressions and source language feature structures. Syntactic analysis is performed (1106) on the source language feature structures and entries of a bilingual example database to determine a pragmatic type of the source language feature structure and a syntactic type of sub-structures of the source language feature structures. A fast match is performed (1108) among the sub-structures and the example database to determine compatibility between the sub-structures and entries of the example database. A best match (1110) is performed among the sub-structures and the example database to optimize matches between the sub-structures and the example database entries. Target language feature structures are generated (1114) using selected entries of the example database, and target language expressions are generated comprising the target language feature structures. A grammatically correct speech output is provided comprising the target language expressions. |
