2001 Fiscal Year Final Research Report Summary
Research on analysis and understanding of natural language action commands
| Project/Area Number |
12308014
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Japan Advanced Institute of Science and Technology (2001)
Tokyo Institute of Technology (2000) |
|---|
Principal Investigator |
SHIRAI Kiyoaki Japan Advanced Institute of Science and Technology School of Information and Science Associate Prof., 情報科学研究科, 助教授 (30302970)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
MOCHIZUKI Hajime Japan Advanced Institute of Science and Technology School of Information and Science Associate, 情報科学研究科, 助手 (70313707)
TOKUNAGA Takenobu Japan Advanced Institute of Science and Technology Graduate school of Information Science and Technology, Tokyo Institute of Technology Associate Prof., 大学院・情報理工学研究科, 助教授 (20197875)
SHIMAZU Akira Japan Advanced Institute of Science and Technology School of Information and Science Prof., 情報科学研究科, 教授 (60293388)
|
|---|
| Project Period (FY) |
2000 – 2001
|
| Keywords | animated agent / morphological analysis / syntactic analysis / anaphora resolution / ellipses resolution / man-machine interface |
|---|
| Research Abstract |
This research project aims at development of a man-mchine interface Using natural language and construction of an animated agent system. Animated agent is an intelligent agent realized in a virtual environment, whioh can understand and follow commands given by natural language.
We have developed fundamental techniques to construct such an animated agent. First, we have improved the module of morphological any syntactic analysis, Especially, we have generalized it so that the module could be applicable any applications. Next, we have developed techniques to analyze ill-formed sentences using semantic roles of words, because iil-formed sentences were frequently uttered in speech, Furthermore, several elements is omitted when a human gives action commands by natural language. In order to tackle this problem, new algorithm ofanaphora resolution and ellipses resolution hive also been developed.
Finally, we constructed the animated agent system KAIRAI by integrating the above fundamental techniques. Software agents in KAIRAI can understand and follow commands given by humans, although.commands are limited. KAIRAI is a public software on the internet, and any related reseatchers can use this system.
|
