| IT operation and maintenance system is a software platform, which is to run and maintain the operating ambient of information system and business system. IT operation and maintenance system has been implemented in Northwest China Grid Company limited, and a large number of operation and maintenance events are recorded in this system, every event have class attributes called Cclass and detailed description attributes called Discription. The text clustering method is used to analysis these operation and maintenance events in this paper, to confirm the similar events and methods. The new operation and maintenance event is solved. The main research work are as follows:In chapter one, the data of operation and maintenance events are text type. The data of Cclass is pre-processed and the data type is changed to numerical type which is suitable to the algorithm.In chapter two, the BIRCH algorithm and its improved algorithm called BC-BIRCH are studied, thus a defect that the insert new data only can match with the nearest cluster is avoided. Then the procedures of BIRCH algorithm and BC-BIRCH algorithm based on java language have been given, and the cluster to Cclass’ s properties has been achieved. In the cluster of new operational events, its similar events according to the keywords of Discription’s properties has been determined.In chapter three, the K-means algorithm and its improved algorithm Add-K-means are studied, and the efficiency of the algorithm are improved by joining the new data processing method based on the original K-means algorithm. Then the procedures of K-means algorithm and Add Data-K-means algorithm by using the java language have been achieved, and the cluster to the Cclass’ s properties has been achieved by putting the number of the cluster of the BIRCH algorithm as the k value of the K-means algorithm. In the cluster of new operational events, its similar events according to the keywords of Discription’s properties has been confirmed. |
PDF Full Text Request