Research Of Textual Periodicity Data Mining In Temporal Data

The time of the information is an important attribute in the life, and the information in the database always has the time dimension. Temporal data mining has become an active research area. More and more attention is paid to the discovery of periodicity in temporal data. Periodic pattern is a particularly interesting feature that could be used for understanding the temporal data and reflecting the evolution rules. At the same time, most of the information in the life exists in the form of text. However, little attention has been paid on the textual data contain with time-property. It will be a very interesting research in the textual periodicity data mining.We study the problem of textual periodicity detection in temporal data as follows:First, we describe the research background of text mining and temporal data mining, and analyze the inadequacy of textual periodicity detection. Then we address the contents of this paper.Second, we define the strictly mathematic notion of multi-granularity time interval and granularity conversion on the basis of temporal type, temporal factor and time granularity. Then, a periodic pattern of textual data with multi-granularity time is proposed.Third, in the base of the multi-granularity time interval, we formally define a strict periodic pattern which the length of the pattern is fixed, and a loose periodic pattern which the length of the pattern can be altered in a range. At the same time, we study the support and the confidence of the periodic pattern. Then we prove the relevant property.Finally, we give the corresponding mining algorithm based on the Apriori algorithm. We use the algorithm references some thought of ant colonies algorithm to calculate the length of the periodic pattern. By testing virus textual data, the proposed algorithm shows that some efficient periodic patterns are obtained.The main achievements in this paper contain the following results:1. We define strictly mathematic notion of multi-granularity time interval and granularity conversion. Then we prove some of their properties.2. We proposed the periodic pattern of textual data with multi-granularity time.3We define the notion of strict periodic pattern and loose periodic pattern. At last we present the algorithm references some thought of ant colonies algorithm to mining the periodic pattern of the textual data.