期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2014
卷号：5
期号：3
页码：4541-4545
出版社：TechScience Publications
摘要：Classification of computer users is very useful for assisting them, anticipating their future actions. In addition, it is very useful for making recommendations to a user based on the histories of other users with similar preferences, detecting changes in the behavior of a user, and so on. Several approaches for classifying users are available, however many of them do not consider the changes in user’s behavior, as it is essential in some of the categories of users. For example, a computer user behavior is represented as the sequence of commands issued during various sessions. In such cases, the user behavior is not necessarily fixed but rather it changes, it is necessary to consider his evolving nature. Proposed work deals with Prototype Based approach to correctly classify the created profiles. Although there are different strategies are available for generating prototype, it is necessary to investigate effectiveness of statistical distance metrics for prototype creation. The work presented in this paper deals with selection of the best statistical distance metrics for prototype generation. It can be applicable to any environment where user behavior is represented as sequence of actions or events.
关键词：Behavior Recognition; Sample density; Sequence;Learning; Prototype Based Classifier; Statistical Distance;Metrics
