2000 Fiscal Year Final Research Report Summary
A Knowledge Revision Method based on Similarity Observations
| Project/Area Number |
11680375
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Hokkaido University |
|---|
Principal Investigator |
HARAGUCHI Makoto Hokkaido University Graduate School of Engineering, Professor, 大学院・工学研究科, 教授 (40128450)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KAKUTA Tokuyasu Graduate School of Law, Lecturer, 大学院・法学研究科, 講師 (80292001)
SADOHARA Ken Electrotechnical Laboratory, Researcher, 知能情報部, 研究員
OKUBO Yoshiaki Hokkaido University, Graduate School of Engineering, Instructor, 大学院・工学研究科, 助手 (40271639)
|
|---|
| Project Period (FY) |
1999 – 2000
|
| Keywords | Knowledge revision / Similarity between concept / Abstraction-based similarity detection |
|---|
| Research Abstract |
We proposes a new framework of knowledge revision, called Similarity-Driven Knowledge Revision.
In our framework, the revision is invoked based on a similarity observation by users and is intended to match with the observation.
Particularly, we are concerned with a revision strategy according to which an inadequate typing in describing an object-oriented knowledge base is corrected and revised by specializing the inadequate types to more specific ones without loss of original inference power.
That is, a minimal revision can be achieved according to our revision framework.
In order to realize it, we introduce a notion of{\em extended sorts}.
An extended sort can be viewed as a concept that does not appear explicitly in the original knowledge base.
If a variable typing with some sort is considered too general in the original knowledge base, the typing is modified by replacing the general sort with more specific extended sort.
Such an extended sort can be identified by forward reasoning from the original knowledge base.
Especially, a forward reasoning with SOL-deduction is adopted to obtain extended sorts efficiently.
Some experimental results show that the use of SOL-deduction can drastically improve the computational efficiency.
|
