Research on Association Rules Mining for Data Stream

Qingfeng Li ； Wengfeng Peng

《International Core Journal of Engineering》 8卷2期 (2022/02) Pp. 218-225

https://doi.org/10.6919/ICJE.202202_8(2).0027

摘要

As data stream becomes common in many modern systems, data stream mining has gained its importance in recent years. Since traditional static data mining techniques is not sufficient for analyzing this type of new data, developing new mining algorithms becomes an emergent need. In this paper, we apply Frequent Tree Interpolation Method (FTIM) to mine association rules. It differs from traditional FP-TREE algorithms in two aspects. First, FTLM uses ascending instead of descending order to create frequent tree. As a result, the searching method for mining frequent itemsets evolves from the conditional search in FP-TREE to unconditional search, which saves computation time. Second, interpolation is applied to construct the frequent tree. If an existing item is included in the new input transaction, FTIM inserts the transaction into the frequent tree of this item, rather than to create a new branch. This decreases the branches of the frequent tree, and reduce the memory space required for constructing the data stream frequent tree. Experiments show that the FTIM algorithm outperforms the traditional FP-TREE algorithm in both speed and scalability.

關鍵字

Data Stream ； Association Rules ； FTIM Algorithm ； FP-TREE Algorithm

參考文獻

Agrawal R,Imielinski T,Swami A.Mining association rules between sets of items in large databases.In:Proceedings of ACMSIGMOD International Conference on Management of Date, Washington DC, 1993.207~216.

PAN M,HUANG X J,HE T T ,et al. A Simple Kemel Co-occu-rrence-Based Enhancement for Pseudo-Relevance Feedback. Journal of the Association for Information Science and Technology, 2020, 71(3):264-281.

BOUZIRI A, LATIRI C,GAUSSIER E. LTR-Expand:Query Expansion Model Based on Leaming to Rank Association Rules. Journal of Inteligent Information Systems,2020,55:261-286.

Han J,Jian P et al.Mining frequent patterns without candidategeneration.In:Proceedings of ACM SIGMOD International Conference on Management of Data,Dallas,TX,2000.1~12.

Cheung D W,Lee S D,Kao B.A general incremental techniquefor maintaining discovered association rules. In: Proceedings ofdatabases systems for advanced applications,Melbourne,Australia,2007.185~194.