International Journal of Data Mining, Modelling and Management

Shinde, Subhash K. ; Kulkarni, Uday V.

doi: 10.1504/IJDMMM.2010.035561pmid: N/A

Recommendation systems represent user preferences for the purpose of suggesting items to purchase or examine. They have become important applications in electronic commerce for information access and for providing suggestions that effectively prune large information spaces so that users are directed toward those items that best meet their needs and preferences. A variety of techniques have been proposed for performing recommendation, including content, collaborative and knowledge-based techniques. However, there remain many challenges in deploying traditional recommendation techniques for e-commerce. This paper addresses these key challenges and proposes new techniques that combine the content and collaborative-based filtering to capitalise on their respective strengths and thereby achieve better performance. We describe new architecture for hybrid recommendation system. The results obtained empirically demonstrate that the proposed recommendation algorithms perform better and alleviate the challenges such as data sparsity and scalability.

Li, Shuxin ; Lang, Sheau-Dong

doi: 10.1504/IJDMMM.2010.035562pmid: N/A

The problem of finding highly correlated pairs is to output all item pairs whose (Pearson) correlation coefficients are greater than a user-specified correlation threshold. Effective discovery of such item pairs is of primary importance in many real data mining applications. Algorithm and Taper algorithm are special cases of our new algorithm with respect to the number of segments. Experimental results on real datasets demonstrate the feasibility and superiority of our algorithm. Recently, the Taper algorithm is developed to discover the set of highly correlated item pairs. In this paper, we present a generalised Taper algorithm to find strongly correlated pairs between items by partitioning the collection of transactions into different segments, so as to achieve better pruning effect and less running time. Consequently, it can be proved that both are naive.

Yu, Liang ; Gao, Lin ; Wang, Danyang ; Fu, Shaofeng

doi: 10.1504/IJDMMM.2010.035563pmid: N/A

Detecting community structure is a powerful approach to understanding complex networks. Recently, modularity function Q has been widely used as a measure to identify communities in complex networks. However, optimising Q function has some resolution limitations. In this paper, we present a new quantitative function DQ (degree modularity) that detects community structure based on local connectivity of communities. We first prove that the function DQ can improve the resolution limitations of modularity Q. Furthermore, we experimentally evaluate the performance of the new quantitative function using a variety of real and computer-generated networks and find communities of widely differing sizes can be detected with higher sensitivity and reliability. Also, even in large-scale biological networks, such as protein-protein interaction (PPI) networks, we can obtain higher matching rate between the predicted protein modules and the known protein complexes. All the experimental results support the usefulness of the new quantitative function DQ as the measure for community structure detection.

Ramdane, Chafika ; Meshoul, Souham ; Batouche, Mohamed ; Kholladi, Mohamed-Khireddine

doi: 10.1504/IJDMMM.2010.035564pmid: N/A

The emerging field of quantum computing has recently created much interest in the computer science community due to the new concepts it suggests to store and process data. In this paper, we explore some of these concepts to cope with the data clustering problem. Data clustering is a key task for most fields like data mining and pattern recognition. It aims to discover cohesive groups in large datasets. In our work, we cast this problem as an optimisation process and we describe a novel framework, which relies on a quantum representation to encode the search space and a quantum evolutionary search strategy to optimise a quality measure in quest of a good partitioning of the dataset. Results on both synthetic and real data are very promising and show the ability of the method to identify valid clusters and also its effectiveness comparing to other evolutionary algorithms.

Vijaya, Kalavakonda ; Khanna Nehemiah, H. ; Kannan, A. ; Bhuvaneswari, N.G.

doi: 10.1504/IJDMMM.2010.035565pmid: N/A

In this paper, we have proposed a medical diagnosis system for predicting the severity of the cardiovascular diseases. The system is built by combining the relative advantages of fuzzy logic, neural network and genetic algorithm. The input variables that are non-discrete are fuzzified and fed as input to train the neural network. The neural network is trained using a genetic algorithm and used to identify the fuzzy rules that are significant for the purpose of classification. The rules identified by the neural network are further pruned and stored in the knowledge base. The rules in the knowledge base are used by inference and forecasting subsystem to predict the severity of the disease, for a given set of input data. Using the proposed approach, we have obtained classification accuracy of 88.35%.

Zhang, Yuejin ; Zhang, LingLing ; Liu, Ying ; Shi, Yong

doi: 10.1504/IJDMMM.2010.035566pmid: N/A

Association rule mining generates large quantities of rules, but not all of them are useful for decision making. In order to find the genuine useful knowledge for decision making, we propose an intelligent knowledge discovery model which is a new purpose-oriented approach based on a second order mining from association rules. More specifically, our model consists of two phases. In the first phase, proper objective measures are selected according to the user's goal. In the second phase, we define a new concept of rule utility measure as the subjective evaluation which incorporates user's goal, expert's experience, and domain knowledge. By doing so, the intelligent knowledge, which can support special strategies can be obtained. Experiments on two real world databases validate the effectiveness of our new model.