An Approach to Early Recognition of Web User Tasks by the Surfing Behavior

A study was conducted to investigate Web users' information seeking behavior on online newspapers, distinguishing between the task categories fact finding, information gathering and browsing. Over a period of four weeks, the surfing behavior of 41 users was recorded who additionally kept a diary to document their activities. It was scrutinized whether the surfing behavior shows significant differences depending on the kind of task already at the beginning of an activity, which is a prerequisite for timely reaction to current user needs. According to the results, behavioral aspects, such as the number of pages viewed, scroll and mouse movement behavior etc. produce significant differences already during the first 60 seconds of a task. Nevertheless, classification tests show that these behavioral attributes do not yet lead to a prediction accuracy sufficient for a sound real-time task recognition.