Jiadong Ren

Mining Weighted Closed Sequential Patterns in Large Databases

Previous algorithms mine the complete set of sequential patterns in large database efficiently, but when mining long sequential patterns in dense databases or using low minimum supports, it may produce many redundant patterns and some uninterested patterns. In this paper, a novel weighted closed sequential pattern mining algorithm (WCSpan) is presented, which implements the closed sequential pattern mining with weight constraints, so the uninterested patterns could be pruned and the redundancy could be reduced. This algorithm can find fewer but interested weighted sequential patterns by weighted pruning method and hash structure. The experimental results show that WCSpan algorithm is more efficient than CloSpan and WSpan.

Density-Based Data Streams Clustering over Sliding Windows

Data stream clustering is an important task in data stream mining. In this paper, we propose SDStream, a new method for performing density-based data streams clustering over sliding windows. SDStream adopts CluStream clustering framework. In the online component, the potential core-micro-cluster and outlier micro-cluster structures are introduced to maintain the potential clusters and outliers. They are stored in the form of Exponential Histogram of Cluster Feature (EHCF) in main memory and are maintained by the maintenance of EHCFs. Outdated micro-clusters which need to be deleted are found by the value of t in Temporal Cluster Feature (TCF). In the offline component, the final clusters of arbitrary shape are generated according to all the potential core-micro-clusters maintained online by DBSCAN algorithm. Experimental results show that SDStream which can generate clusters of arbitrary shape has a much higher clustering quality than CluStream which generates spherical clusters.

Efficient Outlier Detection Algorithm for Heterogeneous Data Streams

Data streams outlier mining is an important and active research issue in anomaly detection. Most of the existing outlier detection algorithms can only manipulate numeric attributes or categorical attributes. In this paper, we propose an efficient outlier detection algorithm based on heterogeneous data streams, which partitions the stream in chunks. Then each chunk is clustered and the corresponding clustering results are stored in cluster references. The representation degree and the number of adjacent cluster references of each cluster reference are computed to generate the final outlier references, which include potential outliers. Experimental results show that our approach has higher detection precision and better scalability.

An Approach for Analyzing Infrequent Software Faults Based on Outlier Detection

The fault analysis is critical process in software security system. However, identifying outliers in software faults has not been well addressed. In this paper, we define WCFPOF (weighted closed frequent pattern outlier factor) to measure the complete transactions, and propose a novel approach for detecting closed frequent pattern based outliers. Through discovering and maintaining closed frequent patterns, the outlier measure of each transaction is computed to generate outliers. The outliers are the data that contain relatively less closed frequent itemsets. To describe the reasons why detected outlier transactions are infrequent, the contradictive closed frequent patterns for each outlier are figured out. Experimental results show that our algorithm has shorter time consumption and better scalability.

Frequent Itemset Mining Based on Bit-Sequence

Most existing frequent itemset mining algorithms based on bit-sequence will generate many candidate itemsets. In this paper, we present FIM-BS for mining frequent itemset. First we adopt bit-sequence to compress the database and define Ilink-array and FNC-tree, frequent item and related pointer information are saved in Ilink-array. Then we insert itemsets into FNC-tree according to the pointers, the nodes of FNC-tree contain itemset, itemset’s count and pointers. (K-1)-itemset subsets are inserted while we insert a new K-itemset to FNC-tree, and the pointer of the K-itemset points to (K-1)-itemset nodes in order to insert K-itemset’s superset or increase the count quickly. This algorithm uses top-down traversal strategy and Apriori property to get mining result quickly. Experimental results also show that the proposed FIM-BS algorithm can reduce the cost of time and improve mining efficiency.

An Improved OLAP Join and Aggregate Algorithm Based on Dimension Hierarchy

The OLAP (online analytical processing) queries are always involved with queries on the massive dataset. As a result, how to perform multi-table join and aggregate operations becomes the key issue. A Join and Aggregate Algorithm Based on Dimension Hierarchy (JABDH) is proposed in this paper. Considering the semantic characteristic which is not in all the dimension hierarchies, dimension hierarchical encoding is used to retrieve the matching dimension hierarchies and evaluate the set of query ranges for semantic dimension hierarchies. To improve the efficiency of multi-table join and aggregate operations for non-semantic dimensional hierarchies, join and aggregate operations are translated into bitmapped join index of fact table. The performance analysis and experimental results show that JABDH has improved the speed of queries and the efficiency of the OLAP queries.

Hierarchical K-means clustering algorithm based on silhouette and entropy

Hierarchical K-means clustering is one of important clustering task in data mining. In order to address the problem that the time complexity of the existing HK algorithms is high and most of algorithms are sensitive to noise, a hierarchical K-means clustering algorithm based on silhouette and entropy(HKSE) is put forward. In HKSE, the optimal cluster number is obtained through calculating the improved silhouette of the dataset to be clustered, so that time complexity can be reduced from O(n2) to O(k × n). Entropy is introduced in the hierarchical clustering phase as the similarity measurement avoiding distance calculation in order to reduce outlier effect on the cluster quality. In the post processing phase, the outlier cluster is identified by computing the weighted distance between clusters. Experiment results show that HKSE is efficient in reducing time complexity and sensitivity to noise.

A New Method of Software Security Checking Based on Similar Feature Tree

In order to optimize the fault feature database(FFD) and to improve the checking efficiency of software fault, in this paper, a novel method of software security checking based on similar feature tree(SFT) is proposed. All of fault feature patterns in FFD are considered as nodes of SFT. SFT is a special binary tree in which the left child of each node is a super-pattern of the node and the right child is a brother-pattern of the node. An improved K-modes clustering algorithm and association rules are used to construct SFT. According to the characteristics of association rules, if a fault feature which is obtained by program slicing from software procedure cant successfully matches to root of SFT, then it matches to the right child of the root. Otherwise the fault feature matches to the left child of the root. This process is iterated according to even left not right rule until leaf node in right sub-tree or no node can be successfully matched in left sub-tree. Finally, the checking result is given by SFT. Experimental results show that our method has higher efficiency of software fault checking.