Discovery of Personalized Information Systems Usage Patterns

Users use information systems to accomplish their tasks usually consisting of multiple steps. Each user or user group might have a preferred sequence of the steps. Adaptive information systems attempt to exploit such usage patterns, and it is expected that adaptivity can be improved through personalizing processes and accounting for temporal and sequential dependencies. However, there is a lack of empirical evidence of existence of personalized information systems usage patterns. The objective of this paper is to analyze empirical information systems usage data in order to confirm existence of personalized information systems usage patterns, which could be further used in developing adaptive information systems. Empirical data analyzed are derived from log files of customers service website of a telecommunication company and of university’s e-learning system. The Longest Common Subsequence algorithm is used to discover patterns in task execution. It is found that frequency of patterns observation varies significantly between the data sets although in both cases personalized patterns are observed more frequently than general patterns. The personalized patterns also are more precise than the general patterns