Thesis Research Project Simon Fuchs
Context-data driven classification of support incidents to improve a support ticket system
Aim: I optimize the UCC support desk using context-data, including the development of several it artifacts for better incident creation, context-data driven automated ticket classification and NLP-based semi-automated answering of support tickets.
Abstract: Movement from Software-as-a-Product (SaaP) to Software-as-a-Service (SaaS) with an increasing rate of updates, interface changes and newly added functions to software products run in the cloud leads to increasing volumes of support incidents raised by customers at SaaS providers. Therefore, the economic pressure of automating service desks increases constantly. Machine Learning algorithms offer a new range of promising approaches to automate such services. Against this background, the dissertation aims to analyze the present state-of-the-art in the field and to derive requirements to modern Support Ticket Systems. Furthermore, we aim to develop a semi-automated Support Ticket System based on these requirements. The system shall be able to automatically distribute tickets; collect context data; and to use these to help a customer create a ticket. Further, it shall be able to help process the issue by proposing answer texts and providing context information.
The advantages of such a proposed IT-artifact for businesses are apparent. It would reduce manual labor in a service desk significantly, either by reducing labor expense for a company or by relieving working force that could be deployed in sectors that are more productive. We expect to develop a fully operating, semi-automated service desk artifact for the UCC support and to optimize, reorganize and simplify the management and processing of support incidents at the UCC support.