Mining top issues from contact center logs for self help portals

Abstract

Self help portals are increasingly popular means for enabling users to find information, resolve problems, and process transactions directly without calling contact centers. Such portals can result in faster problem resolution for users and cost savings for the contact center. However, the effectiveness of self help portals is often limited by the topicality, recency, and relevance of the knowledge (documents, etc.) that users are provided access to. In this paper, we present a system and an architecture for automatically mining the top issues - questions people are calling contact centers for - and presenting corresponding solution documents to users through self help portals. Running the top issues mining regularly (e.g. hourly, daily, etc.) ensures dynamically updated and relevant content on the portals and can greatly reduce costs at contact centers by avoiding calls. Furthermore, top issues mining can highlight the knowledge gap or the issues for which no solution documents currently exist. We describe our end to end system, present algorithms for mining, and discuss the knowledge gap that can prevent self enabling portals from realizing its potential benefits. © 2008 IEEE.