Towards Developing Oops! Solver for Elderly Care

Abstract

Long-term care costs, burdens of caregivers, and their resultant shortage are prevalent concerns in an increasingly aging society. Use of information and communication technologies is one possible solution to expedite human resource development; however, there are few knowledge resources that are interpretable by computers. In this study, we present Example of structured manuals for elderly care as a knowledge resource for question-answering systems. We conducted semi-structured interviews with caregivers in a care facility and retrieved information from an open data set to collect 'Oops' incidents in daily caregiving. Based on the collected incidents, we created questions worded in natural language and corresponding computer-interpretable queries. Consequently, we obtained 150 Oops incidents and 33 computer-interpretable queries and confirmed the queries can retrieve relevant knowledge from the knowledge resource.