Identifying Influential Users in On-line Support Forums using Topical Expertise and Social Network Analysis

Abstract

On-line support forums are a common method for businesses to provide product support for customers. In addition to trouble-shooting and how-to guides, on-line forums also serve the important purpose of allowing customers to interact and discuss the business's products. These interaction are an important factor in influencing customer opinions, and subsequently the adoption and use of products and services. The identification of influential users on these forums would therefore enable businesses to more effectively disseminate information and communicate with customers. In this paper we develop a method for identifying influential users in support forums using topical expertise and social network analysis. One of the key challenges when analyzing influence in this context is that the users are generally less socially active than users on other social networks such as Twitter and Facebook. In order to address this issue we have taken a broader view of a social network and considered all of the users that a particular user has interacted with instead of just the subset of users for which there is an explicit relationship. The user's expertise in a particular category is then used to determine the weight or influence of each individual interaction. Finally, the influence of the top influential users is then categorized as positive or negative based on sentiment analysis of their posts. We have demonstrated our approach using data from the Cisco Support Community, an on-line question and answer forum for computer networking products and services. Preliminary results show that the method was able to identify influential users in a discussion forum for network security products.