| Co-Investigator(Kenkyū-buntansha) |
AKIBA Tomoyosi AIST, Senior Researcher, 情報処理研究部門, 主任研究員
ITOU Katunobu Nagoya University, Associate Professor, 大学院・情報科学研究科, 助教授 (30356472)
FUJII Atsushi University of Tsukuba, Associate Professor, 図書館情報学系, 助教授 (30302433)
|
|---|
| Research Abstract |
While the number of machine readable information accessible via the World Wide Web is growing, the digital divide problem caused by various barriers, such as keyboard literacy, has become crucial
To counter this problem, we developed a system that can be used via speech information and evaluated its performance by means of experiments
In research on automatic speech recognition and speech dialogue systems, a system with a limited vocabulary was usually applied to restricted domains. However, existing text retrieval systems have been targeted various types of texts, such those on the Web, and therefore the vocabulary size is much larger than that for speech recognition. Thus, we proposed a method to fill the gap between speech recognition and text retrieval in terms of the vocabulary size
Reflecting the rapid growth in utilization of machine readable multilingual texts, including those accessible via the Web, it is feasible that users are interested in retrieving information across languages. However, users who have difficulty formulating foreign queries cannot select and exploit foreign information. To resolve this problem, we integrated multi-lingual and cross-language information retrieval techniques to our retrieval system
Although researchers in the spoken language processing (SLP), information retrieval (IR), and natural language processing (NLP) communities share the target, i.e., human languages, researchers in the different communities have not fully collaborated to improve the quality of Human language technologies in past research project. Consequently, it is difficult for researchers in a community to understand the problems in the other communities. Our project was an example joint research, in which we developed a system including technologies proposed in the SLP, IR, and NLP communities
|
