Research On Log-based Association Rules Analysis Method

With continuous steady development of the Internet industry,new technology constantly emerge,like as artificial intelligence,big data,Internet,mobile payment.These new technologies accelerate depth fusion of economic and social fields,at the same time,also makes the network security faces a bigger challenge.In constructing theory and practice of network security system,log analysis is often combined with firewall technology,intrusion detection technology in data mining,to find useful information in large amounts of data,association rule mining is one important widely used branch of the data mining technology.It has high theoretical value and practical significance that making in-depth research association rules analysis and applying it to the log analysis.In this paper,the related concept of log and the basic knowledge of log association rule mining theory is introduced,and two classical association rule analysis algorithms(Apriori algorithm and FP-growth algorithm)is described by combining examples to analyze their respective defects.A improved scheme based on FP-growth algorithm is proposed,in this scenario,the idea is that scan the transaction database once,form an D-Tree from inserting all transactions detailed,the Tree is a special form of transactional database replicas,contains all the transactions in the transaction database,each transaction corresponding one path in a Tree.When D-Tree set up is completed,the transaction database can be replaced directly.The new algorithm does not need to generate a large number of candidate item sets,but only needs to access the transaction database once,which improves the mining efficiency in both time and space.To verify the effectiveness of the improved scheme,the three algorithms implemented separately,and the experiment results of the three methods are compared.The experimental results show that the new algorithm can excavate faster than the two classical mining algorithms under the condition of ensuring the accuracy of mining.With the increase of data volume,the new algorithm has better performance than the classical algorithm.