<![CDATA[This research explores the analysis of 388 hotel customer reviews to understand guest experiences, employing advanced analytical methodologies to uncover valuable insights for service quality enhancement. Utilizing the Knowledge Discovery in Databases (KDD) framework, the study applies Latent Dirichlet Allocation (LDA) for topic clustering and k-nearest Neighbors (k-NN), enhanced by the Synthetic Minority Over-sampling Technique (SMOTE) for sentiment classification. The integration of these techniques allows for the extraction of coherent thematic patterns and the accurate differentiation of sentiment categories within the reviews. The findings reveal that LDA, evaluated through metrics such as log-likelihood (-54,886.092) and coherence scores (-14.949), effectively captures the underlying themes discussed by guests, providing a clear representation of customer priorities and concerns. Additionally, applying SMOTE significantly improves the k-NN model's performance, achieving an accuracy of 91.43% and a precision of 97.26% by balancing class distributions and enhancing classification accuracy. This approach demonstrates the potential of combining topic modeling and sentiment analysis to derive actionable insights, which can be strategically utilized to optimize service delivery and elevate the overall customer experience in the hospitality industry. The study concludes that leveraging such data-driven methodologies facilitates a deeper understanding of customer feedback, ultimately supporting informed decision-making and continuous service improvement.]]>
Copyrights © 2024
