Modeling of Data Integration and Datamining with Ambiguity

2014 Fiscal Year Final Research Report

• Project/Area Number: 24500182
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Kanagawa Institute of Technology
• Principal Investigator: MATSUMOTO Kazunori 神奈川工科大学, 情報学部, 教授 (40350673)
• Co-Investigator(Kenkyū-buntansha): OHSUGA Akihiko 電気通信大学, 大学院情報システム学研究科, 教授 (90393842)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: データマイニング / データ融合 / 多重世界 / 優先度 / センサーデータ

Outline of Final Research Achievements

This study develops effective methods to prepare useful databases for datamining. It is in general not sufficient to use a database from a single source. We develop three types of methods that depend on different viewpoints. First, we assume a set of background knowledge is attached to the original database. Then a developed method automatically discovers useful data using the Web search, and attaches it to the database. Second, we develop a method to integrate multiple sources of databases into the one of a single source. The point here is to maintain many possibilities of the integration. Third, we develop an automatic acquisition of data with which we estimate the degree of reliability for the original database. These three types of methods and new datamining algorithms with them are implemented, and then the effectiveness is verified through experiments.

Free Research Field

知識工学