|発明の名称||SELF-ORGANIZING MAP FOR REPRESENTING SEMANTIC SPACE|
PROBLEM TO BE SOLVED: To provide a reproducible clustering by overcoming inconsistency of space allocation, which is a drawback of a characteristic map of a self-organizing map method.
SOLUTION: Each unit of the characteristic map of the self organizing map method is allocated in a two dimensional manner by mapping a deflection of a reference vector from a base line to the Y axis and a weight of the reference vector in a semantic space to the X axis so that each unit can be represented separately in a reproducible manner. For one example of calculating the weights in the semantic space, in the case of a plurality of examinations for disease diagnosis, a half of them are allocated to the semantic space of diabetes, over nourishment, and liver disorders, and the remaining half of them are allocated to the semantic space of inflammation, under nourishment/kidney function, and anemia. Then the weights of the vectors in the two semantic spaces are calculated so that the unit related to the diabetes, and the unit related to the liver disorders are displayed separately for example even when being distributed into the other space where a plurality of the meanings are the same.
- 国立大学法人高知大学 地域連携推進センター 知的財産部門
- URL: http://www.ckkc.kochi-u.ac.jp/%7Eckkc0001/chizai/
- Address: 〒780-8073 高知県高知市朝倉本町2-17-47
- TEL: 088-844-8418
- FAX: 088-844-8556