Mining user's Web searching skills through fuzzy cognitive state map

Abstract

The vast majority of college students have been reared as researchers in an environment where boundaries for information have been clearly marked, i.e., that of books and paper text. Increasingly, however, they are called upon to perform tasks in an environment not clearly bounded, that of hyperspace. How do we learn to surf in the unbounded space of hyperlinks or the World Wide Web (WWW)? In order to effectively teach students new surfing skills we must be able to understand how neophyte Web users form the cognitive neurological networks that result in a mental pathway, or cognitive map, that makes more navigable the route to further information as well as the information they set out to find. A Markovian modeling of user behavior is introduced and compared to a fuzzy cognitive map (FCM) that represents the opinions of experts on how users surf the Web. Experts are divided on what causes users to fail their queries on the Web. The paper shows that a viable FCM model can be developed and some limit-cycle equilibria are uncovered. A FCM limit cycle repeats a sequence of events and actions. Limit cycles can reveal cognitive and behavioral patterns of users on the Web. An adaptive FCM is built to reflect its causal behavior in time. This change reflects the user behavior as knowledge of the Web increases with time. The causal behavior learns from data. Users learn new patterns and reinforce old ones.