Project/Area Number |
10838009
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Research Field |
リハビリテーション科学
|
Research Institution | Utsunomiya University |
Principal Investigator |
ADACHI Hisahiro Faculty of Engineering Res.Assoc., 工学部, 助手 (00241844)
|
Project Period (FY) |
1998 – 2000
|
Project Status |
Completed (Fiscal Year 2000)
|
Budget Amount *help |
¥3,400,000 (Direct Cost: ¥3,400,000)
Fiscal Year 2000: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 1999: ¥1,100,000 (Direct Cost: ¥1,100,000)
Fiscal Year 1998: ¥1,300,000 (Direct Cost: ¥1,300,000)
|
Keywords | Sign Language / Machine Translation / Manual Motion Description / News Flash Headline / Similarity / Sign Word / Electronic Dictionary / Paraphrase / 通訳システム / 聴覚障害者 / 自然言語処理 / 要約 / 情報獲得 / 手話アニメーション / ニュース速報 / 手話言語 / 福祉工学 / 機械翻訳システム / 聴覚障害 |
Research Abstract |
To construct example-based Japanese to Sign text translation systems, there is a serious problem that it is difficult to handle ill-formed sentences such that several postpositions are omitted from a Japanes sentence. This project presents a reasoning method for recovering the lost postpositions from ill-formed sentences using a conjunctive table which stored frequencies of the co-occurrence relations of collocations for postpositions and adjacent characters. The frequencies can be computed by training examples. The most appropriate candidate for the recovery is decided by the decision rules derived from the frequencies and conjunctive conditions. This project also describes a synthesis method for sign animations based on verbal descriptions, which is closely related with the organization of an electronic sign dictionary and which has the powerful capabilities for a flexible inverted retrieval and a customization of the sign entries represented as animations. This project shows a prototype system based on this observation, which has a conventional customizing functions and which is available through the internet.
|