An Automatic Case Acquisition Approach

Abstract

As customer satisfaction can be considered as a competitive advantage, technical helpdesk agents have to be able to solve customer’s problems fast and hence need to be very knowledgeable in their particular domain. A Case-Based Reasoning (CBR) system can support them by providing a case base of past problem’s solutions. Building and maintaining a case base, however, is a tedious and costly task. In order to have a good quality, human review is required which can quickly become a bottleneck. How to bestbuild case bases from raw data respectively reuse and leverage hidden knowledge in existing datasets is therefore an ongoing research topic. This study aims to build a case base from a helpdesk ticket dataset in an automatic way, without having a human task bottleneck. As helpdesk tickets are not intended to perform CBR with, the raw data is usually very messy and not properly labelled. Some tickets however can contain valuable troubleshooting information....