Characterizing and Evaluating Users' Information Seeking Behavior in Social Tagging Systems

Social tagging systems in the Web 2.0 era present an innovative information seeking environment succeeding the library and traditional Web. The primary goals of this study were to, in this particular context: (1) identify the general information seeking strategies adopted by users and determine their effectiveness; (2) reveals the characteristics of the users who prefer different strategies; and (3) identify the specific traits of users' information seeking paths and understand factors shaping them. A representative social tagging system, Douban (http://www.douban.com/) was chosen as the research setting in order to generate empirical findings.Based on the mixed methods research design, this study consists of a quantitative phase and a qualitative phase. The former firstly involved a clickstream data analysis of 20 million clickstream records requested from Douban at the footprint, movement, and track levels. Limited to studying physical behavior, it was complemented by an online survey which captured Douban users' background information from various aspects. In the subsequent qualitative phase, a focus group gathered a number of experienced Douban users to help interpret the quantitative results.Major findings of this study show that: (1) the general strategies include encountering, browsing by resource, browsing by tag, browsing by user/group, searching, and monitoring by user/group; (2) while browsing by resource is the most popular strategy, browsing by tag is the most effective one; (3) users preferring different strategies do not have significantly different characteristics; and (4) on users' information seeking paths these exist two resource viewing patterns - continuous and sporadic, and two resource collecting patterns - lagged and instant, and they can be attributed to user, task, and system factors.A model was developed to illustrate the strategic and tactic layers of users' information seeking behavior in social tagging systems. It offers a deep insight into the behavioral changes brought about by this new environment as compared to the Web in general. This model can serve as the theoretical base for designing user-oriented information seeking interfaces for social tagging systems so that the general strategies and specific tactics will be accommodated efficiently