The Research Of The Relationship Analysis System Based On Data Mining

Relationship analysis is to analyze the potential relationships between people from the mass communication data, which can provide decision support to other system. It is easily to dig out the information in the data that applies the technique of data mining to the relationship analysis.The system includes three modules: user interface module, data preparation module and association rules mining module. The user interface module is mainly responsible for receiving the user's parameters and displaying the data mining results. The data preparation module is mainly responsible for data collection and pretreatment. The association rule mining module is primarily responsible for mining patterns.The amount of the data which comes form the relationship analysis system is often large and increasing with time, so this paper adopts fast incremental updating algorithm for association rules which is based on partition algorithm to achieve the system's core module—association rules mining module. Partition algorithm divides the data into several partitions which can process separately. It can not only reduce the memory pressure but also can achieve multi-processor parallel mining and improve the efficiency of the whole system. Fast incremental updating algorithm for association rules can reuse the results which had been mined already but not re-mining on the entire data when the data changes, it can effectively improve the efficiency of the system. Through analyzing the partition algorithm in-depth, we can find that the partition algorithm will produce a large number of invalid candidate item sets in the process of mining which can affect the efficiency of mining. To solve this problem, we improve the partition algorithm with hash technology, and propose the hash-based partition algorithm, and compare the partition algorithm with the hash-based partition algorithm. The experimental results show that the efficiency of the system improved when the algorithm improved compared to previous, the new algorithm can make the system dig out useful patterns in a certain period of time and it has some practical value.