Recently, smart glasses have been applied in the field of maintenance, with which a worker goes well during work by contacting a support operator from a remote location. The worker, however, cannot go well if the support operator is not there. To solve the issues, this paper discusses how a work system should be developed by integrating smart glasses with a question-answering module. We design the work knowledge acquired from skilled people and manuals as a workflow for when type of question, rule base for how, goal tree for why, domain ontologies, and its RDF data for what. Workers can receive answers from the question-answering module incorporated with multiple knowledge bases for four types of question. The authors conducted a demonstration experiment and found that workers could perform the work more autonomously by importing the system.