Abstract
Guided troubleshooting is an inherent task in the domain of technical supportservices. When a customer experiences an issue with the functioning of atechnical service or a product, an expert user helps guide the customer througha set of steps comprising a troubleshooting procedure. The objective is toidentify the source of the problem through a set of diagnostic steps andobservations, and arrive at a resolution. Procedures containing these set ofdiagnostic steps and observations in response to different problems are commonartifacts in the body of technical support documentation. The ability to usemachine learning and linguistics to understand and leverage these proceduresfor applications like intelligent chatbots or robotic process automation, iscrucial. Existing research on question answering or intelligent chatbots doesnot look within procedures or deep-understand them. In this paper, we outline asystem for mining procedures from technical support documents. We create modelsfor solving important subproblems like extraction of procedures, identifyingdecision points within procedures, identifying blocks of instructionscorresponding to these decision points and mapping instructions within adecision block. We also release a dataset containing our manual annotations onpublicly available support documents, to promote further research on theproblem.