Research Of Negative Sequential Pattern Mining Technology

Sequential patterns mining is an important research field of Data Mining, essentially, it is aim to mine frequent subsequences (positive sequential patterns) from a sequence database. In short, sequential patterns mining mainly considers positive relationships between elements. Nowadays, a lot of methods have been proposed for mining sequential pattern. The theoretical results of sequential patterns mining have been applied to many commercial areas. In practice, the absence of an element in a sequence may imply significant information. In view of the fact, the research of negative sequential patterns mining has attracted increasing concerns in recent years. So far, very few methods have been proposed to mine negative sequential pattern. Furthermore, even there is not a unified definition about negative containment. And all of these negative sequential mining algorithms which have been proposed by experts exist some problems should be solved. For example, the mining efficiency of these algorithms isn't high and so on. And in the mining process of these negative sequential patterns mining algorithms, based on the definition about negative containment, the supports of these candidate negative sequences are counted by a scanning of the database to determine the frequent negative sequences. But the method is undoubtedly inefficient. Therefore, this paper discusses this issue and proposes a new negative sequential pattern mining algorithm which based on direct calculating support method. And ours experiment can verify effectiveness to the new negative sequential pattern mining algorithm.In addition, this paper introduces concepts and technologies of data mining, sequential pattern mining and negative sequential pattern mining. This paper describes the key technology of sequential pattern mining and negative sequential pattern mining and then discusses these existing definitions of negative containment, does experiments using the same dataset, and compares the results.