Multi-Relational Frequent Pattern Mining Algorithm And Its Application Research

Most of the current application data is usually stored in a relational database of multiple relationships. Multi-Relational Data Mining is to find model and rules from the database. The mining results can be applied to the recommendation system. Recommendation system is one of the main ways to solve the Internet information overload, has been widely used in e-commerce and other fields.Frequent pattern discovery is one of the important tasks of data mining. Most frequent pattern mining algorithms can only be found in a relationship, if you want to multiple relational data mining, there are many problems. In the current data mining research, Multi-Relational data mining is one of the important areas of rapid development. Multi-Relational frequent pattern discovery method can directly from the multiple relations mining complex frequent pattern, to avoid the limitation of single relation model method. This paper proposes a Multi-Relational High Utility mining algorithm, is based on the star schema. This algorithm gives the different utility value for each item, to mining the most valuable frequent pattern, The application of this algorithm have great practical meaning.As considering users’accessing order, recommendation approach based on sequential pattern is becoming one hot topic in the field of recommender system. To improve the level of personalization, a recommender algorithm named PRWSP based on weighted sequential pattern is proposed. A new weighted sequential pattern model is presented at first. In this model, the importance of items to different sequences is considered. Furthermore, by approximation, the rationale of anti-monotonicity in mining weighted sequential pattern is discussed. Thus, the searching space is reduced. Finally, measurement on matching of sequential patterns is defined. Experimental results show that PRWSP is efficient and accurate.Finally, the paper integrated the two algorithms and the corresponding comparison algorithm design and implements a prototype system, and compare the results of the algorithm is analyzed.