Characterizing Web Users Based on Their Request Criteria

Abstract

Understanding the behaviors or characteristics of users surfing a Web site has been playing a crucial role for running a successful website. Early studies of discovering these patterns focused on Web log data. However, when people actually surf a website, they can submit a case online by selecting or filling value in provided fields as preferences or requirements for an object. Interested persons can find this posted case according to these saved criteria. In this paper, an approach for characterizing Web users based on their request criteria is proposed. The goal is to capture the knowledge of Web users' behaviors. Thus, decisions can be made to provide better service or to enhance marketing strategy. This proposed approach consists of three steps: data preprocessing, pattern discovery, and pattern analysis. In the data preprocessing step, useful information is extracted from raw collected request criteria data from a website and transformed to the input format of clustering algorithm. Consequently, users with similar properties of criteria are grouped together. Association rules can be mined among the users with high similarity of request criteria. Finally, the characteristics of users in each group can be inferred or identified easily by analyzing these generated rules. Some marketing strategies, especially for each group's users, are suggested based on their identified characteristics.