The Study Of Results Storage And Application In Data Mining

Data mining is recently core technologies, by which an enterprise realizes commercial intelligence, and it is a key step in knowledge discovery process, which has become a popularly central issue. How to improve the efficiency of data mining has become a popular issue for the academic circle to probe into. Aiming at this issue, the author proposes a new data mining systematic architecture based on the analysis of present system architecture. The main thought is to introduce a management system of pattern base in the original data-mining systematic architecture. This management system of pattern base has three mainly function components:Pattern base-storing patterns of data mining; Management system of pattern base-managing pattern base, through management system of pattern base the users can carry on various operations and management; Monitor-offering automatically trigger mechanism, answering for automatical monitoring changed data and transfering changed data to management system of pattern base, in order to implementing automatical update of patterns by data mining again. So this provides a method for problem of patterns' time-efficiency. This new system improves the efficiency of data mining and shares data-mining results. In order to unify stored data-mining results in pattern warehouse, the author probes intensively into the storage methods of data-mining results (association rules, classification rules, sequential patterns). It proposes the storage methods to use the relational algebra to describe. This realizes consistent storage of data-mining results in relational database. As a result of extensive application of relational database, this improved storage methods provide unified platform for sharing data-mining results in different mining systems. Based on the above work, the author carries out researches to the operation of data-mining results, and then it puts forward SPQL (Structured Pattern Query Language). SPQL Sentences will be used to realize the management and query of data-mining results. The author gives out the realized methods of SPQL sentences in relational database, namely it uses SQL sentences to explain SPQL sentences. This query language makes things convenient for operating data-mining results stored in pattern warehouse, which provides an effective method for managing and sharing data-mining results. The author also gives out the methods to apply SPQL sentences in synthetic management system of Medicine Co. Ltd, and developed an experimental system. This system has proved that data-mining results storage method is feasible and correct.