Applying KDD(Knowledge Discovery in Databases) technology in data analysis is helpful to improve data-analysis ability and productivity in photosensitive materials enterprises. Based on the character of data and knowledge of the field, contrastive analysis in normal KDD technology is expounded in this paper. The architecture of DAES (Data Analysis Expert System) in photosensitive materials field is studied, and a knowledge–describing approach proper to the field is represented. A data-analysis model system Based on BP ANNs(Artificial Neural Networks) technology is developed. Three BP knowledge models, which reflect relations between emulsion quality parameters and raw materials parameters are given based on the system. And a new KDD approach Based on BP ANNs and Nearest-Neighbor Approach is given and tested in emulsion data-analysis.
|