Summarization of Feedback from Residents in Urban Area Using the Unsupervised Method

In light of the rapid growth of urbanization in Malaysia, many people have decided to congregate to the cities to gain a better quality of life as what believed. However, it is not as expected when different problems arise daily. The residents’ voices are being ignored, and the same urban problems keep happening even though there are complaints everywhere, including on the social media. To cast light on this issue, the current paper attempts to summarize the residents’ feedback using the unsupervised method in the Data Mining approach. The residents’ feedback or dataset were collected from Twitter and CARI Infonet, which is a total of 2320. Moreover, Latent Dirichlet Allocation (LDA) method is selected to perform Topic Modelling. To extract noteworthy topics in the dataset, the Coherence Score measure is performed to find the optimal number of k-values. Finally, three topics were identified and clustered according to their similarity of words: “road problems and traffic congestion”, “public transport”, and “pollution.” The results provide insightful information to the stakeholders, particularly urban policymakers, to lead them to a strategic planning decision-making process reflecting urban residents’ desires.