Abstract | ||
---|---|---|
Toshiba participated in the Japanese monolingual track and the English-Japanese cross-language track at NTCIR-3, and achieved the highest retrieval perfor- mances among the DESCRIPTION runs. Our pre- submission experiments using the KIDS system fo- cused primarily on effective Japanese monolingual retrieval, and addressed several problems including term selection enhancement and ranked output com- bination. Moreover, our post-submission reruns using the recently developed BRIDJE system substantially outperformed our official runs. In addition, our post- submission experiments include a new cross-language run which used the internal translation data of our |
Year | Venue | Keywords |
---|---|---|
2002 | NTCIR | machine translation.,pseudo-relevance feed- back,bridje,ranked output combination,machine translation system,which further improved performance on average. keywords: kids,machine translation |
Field | DocType | Citations |
Ranking,Speech recognition,Presubmission,Natural language processing,Artificial intelligence,Engineering | Conference | 8 |
PageRank | References | Authors |
0.73 | 1 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Tetsuya Sakai | 1 | 1460 | 139.97 |
Makoto Koyama | 2 | 83 | 8.49 |
Masaru Suzuki | 3 | 18 | 2.58 |
Toshihiko Manabe | 4 | 90 | 8.90 |