Search

Mining high dimensional data is an urgent problem of great practical importance. Although some data mining models such as frequent patterns and clusters have been proven to be very successful for analyzing very large data sets, they have some limitations. Frequent patterns are inadequate to describe the quantitative correlations among nominal members. Traditional cluster models ignore distances of some pairs of members, so a pair of members in one big cluster may be far away. As a combination and complementary of both techniques, we propose the Maximal-Correlated-Member-Cluster (MCMC) model in this paper. The MCMC model is based on a statistical measure reflecting the relationship of nominal variables, and every pair of members in one cluster satisfy unified constraints. Moreover, in order to improve algorithm's efficiency, we introduce pruning techniques to reduce the search space. In the first phase, a Tri-correlation inequation is used to eliminate unrelated member pairs, and in the second phase, an Inverse-Order-Enumeration-Tree (IOET) method is designed to share common computations. Experiments over both synthetic datasets and real life datasets are performed to examine our algorithm's performance. The results show that our algorithm has much higher efficiency than the naïve algorithm, and this model can discover meaningful correlated patterns in high dimensional database. [ABSTRACT FROM AUTHOR]

Current sequential pattern mining algorithms often produce a large number of patterns. It is difficult for a user to explore in so many patterns and get a global view of the patterns and the underlying data. In this paper, we examine the problem of how to compress a set of sequential patterns using only K SP-Features(Sequential Pattern Features). A novel similarity measure is proposed for clustering SP-Features and an effective SP-Feature combination method is designed. We also present an efficient algorithm, called CSP(Compressing Sequential Patterns) to mine compressed sequential patterns based on the hierarchical clustering framework. A thorough experimental study with both real and synthetic datasets shows that CSP can compress sequential patterns effectively. [ABSTRACT FROM AUTHOR]

In this paper, we propose a method called CCWrapper (Classification-Cluster) to extract target data items from web pages under the guide of the predefined schema. CCWrapper extracts and combines the different HTML nodes features, including the style, structure, thesaurus and data type attributes into one unified model, and generates the extraction rules with Bayes classification in the training step. When the new HTML page is handled, CCWrapper generates the probability of the target element for each HTML node and clusters the HTML nodes for extraction based on the intra-document relationship in the HTML document tree. The preliminary experimental results on real-life web sites demonstrate CCWrapper is a promising extraction method. [ABSTRACT FROM AUTHOR]

This paper proposes a new method, KCAM, to measure the structural similarity of XML fragments satisfying given keywords. Its name is derived directly after the key structure in this method, Keyword Common Ancestor Matrix. One KCAM for one XML fragment is a k × k upper triangle matrix. Each element ai,j stores the level information of the SLCA (Smallest Lowest Common Ancestor) node corresponding to the keywords ki, kj. The matrix distance between KCAMs, denoted as KDist(·,·), can be used as the approximate structural similarity. KCAM is independent of label information in fragments. It is powerful to distinguish the structural difference between XML fragments. [ABSTRACT FROM AUTHOR]

Unlike Internet portals, a typical Web Information System contains lots of user operations. However, existing Web design frameworks focus exclusively on data presentation: the process of user operations is still achieved through low-level programming. That makes WIS hard to be constructed or maintained. In a Web Information System auto-construction Environment (WISE), the translation of user operations to concrete data processing code in persistent data sources is explored. This paper discusses the semantics of typical operations. The implementation code of those operations can be auto-generated with the aid of the mapping from template attributes to persistent objects. [ABSTRACT FROM AUTHOR]

This paper discusses a method for dynamic web service composition, which is based on Markov Decision Processes (MDP). It is defined on the base of QoS description and addresses the issue of selecting web services for the purpose of their composition. Web service composition patterns including sequential, conditional, parallel and iterative are modeled in MDP. Two computation algorithms for MDP are introduced and implemented. One is backward recursive iteration and the other is forward iteration, both called Value Iteration. Experiments are conducted to evaluate the performance of MDP methods for service composition. Experimental results show that the methods are efficient. The effect of average failure probability of service invocation on performance is also examined in this paper. [ABSTRACT FROM AUTHOR]

System behaviors specify the major functions of domain specific Web Information Systems (WIS). Traditional techniques can not satisfy various requirements or manage innumerous data while developing WIS behaviors. People appeal to a smart tool for implementing the WIS behaviors. This article makes the following contributions: (1) Proposes the concept of domain ontology and behavior ontology to describe the contents and operations of WIS; (2) Extends traditional ECA model to characterize the triggers, parameters, actions as well as validations of WIS behaviors; (3) Analyses the relationships between domain ontology and behavior ontology with rule sets; (4) Implements a tool named WISE Builder with four algorithms to help users building behavior ontology with domain ontology; (5)Shows the feasibility of this technique in a real application case. Keywords: Web Information System, Behavior, Domain, Ontology, Code Generation. [ABSTRACT FROM AUTHOR]

In order to improve privacy preservation and accuracy, we present a new association rule mining scheme based on data distortion. It consists of two steps: First, the original data are distorted by a new randomization method. Then, the mining algorithm is implemented to find frequent itemsets from the distorted data, and generate association rules. With reasonable selection for the random parameters, our scheme can simultaneously provide a higher privacy preserving level to the users and retain a higher accuracy in the mining results. [ABSTRACT FROM AUTHOR]

Replication on geographically distributed, unreliable, P2P interconnecting nodes can offer high data availability and low network latency for replica access. The challenge is how to take good control of the number of replicas and their distribution over well-chosen nodes to get a good replica access performance. We observe that, there exists such a logical node cluster overlay over any P2P data-store's underlying network topology that the replica transmission delay of inter-cluster is much greater than that of intra-cluster because of geographical distance or bandwidth sharing between nodes in different clusters. Based on nodes-clustering, we propose a decentralized algorithm ACB-R to direct the data replication, which can adapt dynamically to the changing replica access patterns or network topologies. The experiment shows that ACB-R can benefit much of the access requests at the price of negligible intra-cluster replica transmission and consequently achieves a good average replica access performance. [ABSTRACT FROM AUTHOR]

Privacy preserving data mining is to discover accurate patterns without precise access to the original data. In this paper, we combine the two strategies of data transform and data hiding to propose a new randomization method, Randomized Response with Partial Hiding (RRPH), for distorting the original data. Then, an effective naive Bayes classifier is presented to predict the class labels for unknown samples according to the distorted data by RRPH. Shown in the analytical and experimental results, our method can obtain significant improvements in terms of privacy, accuracy, and applicability. [ABSTRACT FROM AUTHOR]

The association rule mining is an important topic in recent data mining research. In this paper, a new association rule mining method based on the multiple-dimensional item attributes is proposed through the Market Basket Analysis. The corresponding average weight support is defined, and the AWMAR algorithm is described in detail. Finally, the performance study and results analysis of the improved algorithm is presented. AWMAR algorithm is effective for mining the association rules with acceptable running time. [ABSTRACT FROM AUTHOR]

Privacy preserving data mining is a novel research direction. The main objective is to develop algorithms for modifying the original data in some way, so that the private information remains private even fter the mining process. [ABSTRACT FROM AUTHOR]

This paper proposes a general consistency management methodology within ontology-driven WIS development. The contributions of this paper are follows: 1) introduces the concept structures for representing model ontologies, mappings ontology and their instances. These ontologies and their instances greatly enhance the representation and reasoning capabilities of mapping and also simplify the manipulation 2) Defines and implements a series of general operators to manipulate mapping and model ontologies. 3) Formalizes the consistency management problem and gives an algorithm basing on general operators. 4) Implements a prototype and integrates it with WISE. [ABSTRACT FROM AUTHOR]