| Research Abstract |
The outline of this research is as follows.
1. Development of data-mining tool :
(1) Improvement of data conversion software was created using the LISP language in order to indicate brief the rule drawn. This software sets up one word in a corpus, and attaches a relative index to the word connected with it. Furthermore, a semantic distance and the attribute of directivity are added to them.
(2) It experimented in the data outputted by (1) using Intelligent Miner. Furthermore, the experiment which makes network connection for Intelligent Miner on RX600 (IBM) with IBM PC710 was conducted. The work environment of an experiment has improved by this.
(3) The cascade model aiming at making flexible data mining possible on a propositional logic level was proposed. The data-mining system DISCAS which realizes this was developed.
(4) The data-mining tool using the constellation graph aiming at visual mining was developed. As for this, a display at the 2-dimensional plane of multivariate data and group-ization for data distinction were attained.
2. Application of data-mining tool : Using the corpus data of
(1) EDR English and Japanese, verbal relative node indexation was performed and correlation rule search was performed.
(2) The feature extraction of a literary work was performed and new knowledge was acquired.
(3) Application to chemistry structure data with graph structure was performed, and new knowledge was acquired by the relation between structure and the chemical characteristic.
(4) As application to mental data, new knowledge was acquired about the consciousness structure of the college student of Japan and China, and parents' bringing-up model.
|
