Abstract: The large-scale text data mining is an important branch of the big data analysis and is also a hot research to-pic in recent years．This paper studied algorithm of the textual periodicity data mining with multi-granularity time．First,the concepts of granularity conversion and multi-granularity time interval were presented．Then,a periodic pattern of textual data and an algorithm of the periodic pattern to textual data with multi-granularity time were proposed．Finally,by testing virus textual data,the proposed algorithm shows that some efficient periodic patterns are obtained．The influence of the periodic range on the degree of support and confidence were discussed．This paper provided a new method for the big text data analysis.

Key words: Multiple granularity,Textual data,Data mining,Periodic pattern