The World Wide Web, or the Web, has become the largest and most popular information system to support all types of the activities including commercial campaigns, scientific research, and even illegal activities and terrorism. Unlike most traditional information systems, the structure and contents of the Web represent a considerable amount of latent human annotation and offer us an opportunity to study the behaviors of the Web users and the Web content providers. For example, by examining customers’ online shopping log, we could learn the customers’ preference; by examining terrorist organizations’ Web contents and Web link structure, we could learn the terrorists’ propaganda plans, communication, and cooperation patterns. Such knowledge would be very important for the study of organization behaviors in domains such as e- Commerce, e-Government, and counter-terrorism domains and yet very expensive or difficult to obtain by other means. In this work, we view the Web as a three-layered model and propose a frame work which incorporates theories and methodologies from organizational behavior, computer-mediated communication, human computer interaction, and Web mining domains to study the extraction of hidden knowledge on Web users and content providers’ behavior. We also present three case studies to demonstrate the feasibility and effectiveness of our framework.