Vanhustyöntekijöiden teknologiaan liittämät tunteet – avovastausten analysoiminen tekoälypohjaisen klusteroinnin keinoin
Keywords:
vanhustyö, tunteet, teknologia, luonnollisen kielen käsittely, tekoäly, koneoppiminenAbstract
Technologies employed in the human-centred care work often evoke emotions in workers that affect both the way they work and their encounters with clients. In this article, we explore the emotions elder care workers associate with technology used at work by applying new methods to open-ended data. We ask how the emotions eldercare workers associate with new technology at work, described in natural language, can be grouped using AI-based clustering. In terms of data, we use panel data collected from Finnish elder care workers through an electronic survey in 2019 (N=6903) and 2021 (N=1679). We applied k-means clustering to Transformer-based model based on open-ended response data (N=3806). The results were visualized with word clouds and scatter plots. As a result, we settled on an eight-cluster model. The clusters revealed both the overall structure of the open-ended data and the ways in which care workers responded to open-ended questions. Responses containing positive, negative, neutral, and ambivalent emotions emerged as distinct clusters. In addition, response clusters where emotional experiences were associated with specific characteristics and practices of eldercare work were identified. The results indicate that the tested methods have a lot of application potential in social medicine research. The tested methods can be used to conceptualize poorly understood phenomena and to develop new predetermined closed-ended question for them. Even the small size of data or Finnish language do not pose obstacles.
