AI for Wellbeing : A proof of concept of data mining on wellbeing research trends from Malta

Abstract

Recent years have seen a burgeoning literature on wellbeing measurement and modelling but different countries face different issues at different points in time. Furthermore, in the field of wellbeing, evidence-based policy-making is still at its infancy. This study can be considered as a proof-of-concept to stimulate evidence-based policymaking in any country, at any time. It capitalises on the growing body of research and employs an Artificial Intelligence/Data Mining approach. Relevant publications are acquired from publicly available repositories (such as Semantic Scholar and CrossRef), and from the UM’s Open Access Repository. Topic modelling (using Bert Transformer Models) is applied on the abstracts to identify the main topics, and quantify the association of each publication to the different topics. Each topic is represented by the salient keywords, a dashboard of which is set up to demonstrate popularity over time. Users are able to select any topic on the dashboard to view associated words, the trajectory of that topic’s popularity along the years, a network of the prolific authors within that topic, and how they collaborate. This dashboard will contribute to research on wellbeing and to evidence-based policy making by 1. allowing users to identify the key issues in wellbeing in their particular country (or cluster of countries), 2. Revealing the different levels of engagement of researchers with those themes over time, and identifying under-researched themes 3. Sourcing the publications themselves, by theme and time, and iv. Identifying the experts and their collaborative links.