Author Information
Hock Lin Sng, Agency for Integrated Care, SingaporeJuan Zhen Koh, Synapxe, Singapore
Joycelyn Yun Ting Woo, Synapxe, Singapore
Yu Heng Tan, Agency for Integrated Care, Singapore
Winston Zhao Yang Ma, Agency for Integrated Care, Singapore
Andy Wee An Ta, Synapxe, Singapore
Abstract
The Silver Generation Office (SGO), under the Agency for Integrated Care (AIC), supports seniors in Singapore through home visits to understand their situation and connect them with services to address their needs, if any. Through these visits, valuable qualitative feedback on policies affecting them are collected. This study aims to develop a self-help tool using unsupervised natural language processing to analyse uncategorised free-text feedback to reduce manual effort required in summarising the feedback. A total of 41,891 anonymised and uncategorised free-text feedback collected from April 2022 to March 2023 were analysed using topic modelling algorithm, Non-Negative Matrix Factorisation (NMF), developed on Anaconda JupyterLab. The feedback was analysed by creating a Document-Term Matrix which represents the frequency of terms in each feedback, followed by applying NMF to extract topics with representative keywords. Human evaluation with inter-rater reliability (IRR) assessment was conducted with ten evaluators to assess its accuracy. Results showed that the model achieved over 75% accuracy, with high IRR coefficient above 0.876 after two rounds of evaluation. The model uncovered valuable insights that were previously challenging to obtain through manual efforts. The extracted topics help SGO to better make sense of the data, facilitating sharing of insights with stakeholders to highlight seniors’ needs and preferences which will improve existing policies, programs, and services for seniors. The self-help tool is developed and currently in-pilot, allowing users to automate data preprocessing, conduct textual analysis, and generate visualisation charts. It may potentially enhance SGO’s operational efficiency and reduce man-hours spent on data analysis.
Paper Information
Conference: AGen2025Stream: Aging and Gerontology
This paper is part of the AGen2025 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window
To cite this article:
Sng H., Koh J., Woo J., Tan Y., Ma W., & Ta A. (2025) Data-Driven Approach to Understanding Senior’s Needs: An Automated Unsupervised Learning Solution for Feedback Analysis in Singapore ISSN: 2432-4183 The Asian Conference on Aging & Gerontology 2025: Official Conference Proceedings (pp. 137-156) https://doi.org/10.22492/issn.2432-4183.2025.11
To link to this article: https://doi.org/10.22492/issn.2432-4183.2025.11
Comments
Powered by WP LinkPress