Automation of help desks using case-based reasoning

High quality customer service is an increasingly important requirement of businesses and with growing computer usage and the rapid increase in number of hardware and software products and complexity of these products there has been a proliferation of help desks to handle the influx of queries from users. More intelligent automation is required to share the sought after expertise so that the first line help desk operators can deal with more calls directly and where necessary be advised on referral procedures and any further information that should be sought as the call is being reported. Expertise for problem resolution is sparse and in demand. There is a high turnover of staff. The task involves diagnosis and recovery or problem classification and referral. Diagnosis and classification are both tasks well suited to case-based reasoning (CBR). CBR suits tasks where expertise is built up from previous experience. A rule based system could only share expertise that could be coded in rules. A rule-based system for many help desks would prove impractical, since rules would be needed for all boundary conditions. Input descriptions are typically free-form, incomplete and imprecise. CBR can cope with this sort of information. Help desks typically have plenty of previous data from logged calls, whether these are held on paper or electronically. CBR can cope where similar cases have had different resolutions. A rule-based system would have problems where expertise differs. Knowledge evolves over time. CBR can meet the requirement for knowledge update to handle new problem types.<>