摘要 |
<p>PROBLEM TO BE SOLVED: To exactly and efficiently remove non-related attributes not important for the classification or retrieval of data in a short time while defining data composed of category classified binary attributes as an object. SOLUTION: Concerning a data base 10 expressed in binary attributes al-an and classified into categories C1-Cm, a data number calculating means 2 calculates a total number S of data, number Si of data for each category Ci, number tj of data for each attribute aj to adopt the value of '1' or '0' and number tij of data for each attribute aj to adopt the value of '1' or '0' corresponding to each category Ci. A weight calculating means 4 calculates weight wij for each attribute aj corresponding to each category Ci based on the calculated results of the data number calculating means 2. A distribution calculating means 6 calculates distribution vj of weight for each attribute aj based on the weight wij calculated by the weight calculating means 4. An attribute removing means 8 removes the nonrelated attributes based on the distribution vj calculated by the distribution calculating means 6, a threshold designated by a user or the number of attributes to be removed.</p> |