Hafifah, Hafifah2024-03-012024-03-012023-11PerpustakaanElfitrahttps://repository.unri.ac.id/handle/123456789/11336The people of Pekanbaru can submit complaints to the Pekanbaru City Government through SP4N-LAPOR via the website or application. Based on data from this system, it is evident that the issues faced by the community still occur frequently. This is confirmed by the recurring nature of similar complaints over time, indicating that the issues have not been maximally resolved, and the relevant agencies do not have a deep understanding of the trends in community issues. Therefore, this research will model topics using Latent Dirichlet Allocation (LDA) to group the data into topics that represent the most frequently emerging issues. By understanding these topic trends, the government and relevant agencies can be more responsive in addressing emerging problems. The data used consists of 345 text data obtained from the Pekanbaru City Information and Communication Office. The collected data then undergoes preprocessing stages, including text cleaning, tokenization, normalization, stopword removal, and stemming. After preprocessing, word weighting is performed using Term Frequency-Inverse Document Frequency (TF-IDF). In building LDA, coherence scores are used to determine the most optimal number of topics for topic modeling. Experiments are conducted with 50 and 100 iteration tests. Different numbers of topics, namely 3, 5, and 7, are used for each iteration test. Based on these experiments, the analysis results show that 5 is the most suitable number of topics. The topics identified are, These topics are public services, order, Covid-19, public services, government assistance, and data management.enCoherence ScoreLDAPreprocessingTF-IDFTopic ModellingArticle