A Neural Net Approach to Data Mining: Classification of Users to Aid Information Management
- 181 Downloads
Techniques from the domain of Artificial Intelligence are used increasingly to combat the problem of information overload on the Internet. The vast majority of such techniques and related systems attempt to overcome the problems of information overload by automating the analysis of the content of online documents. In many web-sites (document repositories and e-commerce sites) system logs, documenting user behaviour, are available and can be used as a valuable resource in information management. This information represents a valuable resource to aid in the organisation of information and the presentation of such information to users. In many such systems this information can be represented using a set of tuples indicating which pages/items were visited by a user. Using this information can provide many advantages. A classification of tuples can be used to aid information management and we outline briefly systems which have used the classification algorithm we propose. This paper presents an approach to solving classification problems by combining feature selection and neural networks. The main idea is to use techniques from the field of information theory to select a set of important attributes that can be used to classify tuples. A neural network is trained using these attributes; the neural network is then used to classify tuples. In this paper, we discuss data mining, review common approaches and outline our algorithm. We also present preliminary results obtained against a well-known data collection.
KeywordsData mining neural networks classification web-mining information management
Unable to display preview. Download preview PDF.
- 1.Aggrawal, C.C., Yu, P.S. (1999): Data Mining Techniques for Associations, Clustering and Classification. Proceedings of the 3rd Pacific-Asia Conference on Methodologies for Knowledge Discovery and Data Mining (PAKDD-99).Google Scholar
- 2.Dumais, S. (1991): Improving the Retrieval of Information from External Sources. Behavior Research Methods Instruments and Computers. Vol. 2, No. 23, pp. 229 — 236.Google Scholar
- 3.Mehta, M., Shafer, J., Agrawal, R. (1996): SPRINT: A Scalable Parallel Classifier for Data Mining. Proceedings of the 22nd VLDB Conference.Google Scholar
- 4.Kleinberg, J. (1997): Authoritative Sources in a Hyperlinked Environment. IBM Research Report RJ 10076, May, 1997.Google Scholar
- 5.Lu, FL, Setioni, R., Liu, H. (1995): NeuroRule: A Connectionist Approach to Data Mining. Proceedings of the 21st VLDB Conference.Google Scholar
- 6.Resnick, P., Iacovou, N., Suchak, M., Bergstrom, P., Reidl, J. (1994): GroupLens: An Open Architecture for Collaborative Filtering of NetNews. Proceedings of ACM 1994 Conference on CSCW. pp. 175 — 186.Google Scholar
- 7.Salton, G.A., McGill, M.J. (1983): Introduction to Modern Information Retrieval. McGraw Hill International, 1983.Google Scholar
- 8.Shannon, C. (1948): A Mathematical Theory of Communication. Technical Report, Bell Systems.Google Scholar
- 9.Shardanand, U., Maes, P. (1995): Social Information Filtering: Algorithms for Automating “Word of Mouth”. Computer-Human Interfaces (CHI ‘85).Google Scholar