Mining Domain-specific Component-Action Links for Technical Support Documents

Abstract

IT service support data such as user queries, resolutions done by agents, technical documents are very complex. It is very important to accurately identify the key components and the actions performed on them to better assist applications, such as knowledge extraction and ingestion, search, query, and support chat-bots. Extracting components and correctly linking them to the required actions would also benefit the problem remediation process by drastically reducing the time to resolve and helping the engineers to focus on the steps or section of the document that are relevant and needed to solve the issue. In this paper, we evaluate the performance of four approaches to establish links between mentioned components and actions on technical support documents.