Background

In recent years, there has been a rapid rise in the public policy arena to set subjective well-being (SWB), backed by the system design methodology of well-being-centered design. There are demands from policy-makers to an AI-based research engine to quantitatively detect residents' SWB in their daily lives for visualization as policy evidences

Aims

This research is to develop an AI-based serach engine that infers the degree of SWB from the syntax of free descriptions written by residents on SNS and to visualizes the degree of SWB of residents on a map using the document vectorization function and inference function of AI. 

Methods

By developing the existing natural language processing AI, the authors converted word relationships and semantic expressions in complex document structures into high-dimensional vectors using machine learning. After this document vectorization AI useed the high-dimensional vector, they were converted from the natural language text and the SWB scale from the SWB measurement questionnaire by the same residents who input the natural language text, and it learned the SWB inference by AI model with deep learnings.

Conclusion

 The AI engine developed in this research is in the process of being developed smoothly. We identified multiple linguistic features associated with SWB as vectors.