Cognitive agents for understanding the complexities involved in web-based knowledge-gathering tasks

A Web-based knowledge-gathering task is a common, but complex activity carried out by several users on the Web. The complexity is of two types: One is the inherent complexity involved in the task, which is essentially the information need of the user, and the other is the perceived complexity by the user - that varies according to the proficiency of the user, in the particular subject matter of the task. We present an agent-based cognitive model for understanding both the inherent and perceived complexities. We conducted simulation of our model using logs of user sessions to arrive at the typical complexity values and to qualitatively represent them. We also carried out live experimentation through a prototype system that made estimates of the complexities as users were involved in knowledge gathering tasks, and the results vindicated our estimations.