A Simple Integration of Social Relationship and Text Data for Identifying Potential Customers in Microblogging

Identifying potential customers among a huge number of users in microblogging is a fundamental problem for microblog marketing. One challenge in potential customer detection in microblogging is how to generate an accurate characteristic description for users, i.e., user profile generation. Intuitively, the preference of a user’s friends (i.e., the person followed by the user in microblogging) is of great importance to capture the characteristic of the user. Also, a user’s self-defined tags are often concise and accurate carriers for the user’s interests. In this paper, for identifying potential customers in microblogging, we propose a method to generate user profiles via a simple integration of social relationship and text data. In particular, our proposed method constructs self-defined tag based user profiles by aggregating tags of the users and their friends. We further identify potential customers among users by using text classification techniques. Although this framework is simple, easy to implement and manipulate, it can obtain desirable potential customer detection accuracy. This is illustrated by extensive experiments on datasets derived from Sina Weibo, the most popular microblogging in China.