2005 Fiscal Year Final Research Report Summary
Studies on Database Facility to Support Data Analysis
| Project/Area Number |
15500072
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Media informatics/Database
|
|---|
| Research Institution | KYUSHU UNIVERSITY |
|---|
Principal Investigator |
FURUKAWA Tetsuya Kyushu University, Faculty of Economics, Professor, 経済学研究院, 教授 (00209165)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
MIYANO Eiji Kyushu Institute of Technology, Faculty of Information Engineering, Associate Professor, 情報工学部, 助教授 (10284548)
|
|---|
| Project Period (FY) |
2003 – 2005
|
| Keywords | Database / Data Classification / Classification Hierarchy / Information Retrieval |
|---|
| Research Abstract |
Analysis of collected data is based on Classifying data. This research focused on database facilities to store and supply data for data analysis.
In general, data is classified into tree structures and it is assumed that data belongs to one terminal class (exclusivity and sufficiency). For the variety of data, it is required to prepare classification hierarchies which do not satisfy exclusivity and sufficiency. Data structures for such data and set operations of classes to get the target data to be analyzed were developed. On the other hand, classes are to used to express the ranges of classification and to express the categories of classification. Hierarchical classification structures for these both usages are introduced.
Classification hierarchies are not prepared in advance but grow by generating new classes during data analysis. Facilities to support such analyzing process by classifying data were developed, which enable reorganization of hierarchies and utilization of data on the halfway of analyzing process.
There can be data which should belong to multiple classes, which bring such problems that a class may have data whose semantics is not lower concepts of the semantics of the class and that data are duplicated in several classes. The representation of the duplicated data and the information of data which is not representative solve those problems.
The database facilities developed by this research are useful for organization and utilization of various data to be analyzed.
|
