<p>This study examines the potential of using residents’ sentiments as an explanatory variable for housing price appraisal. We construct a text-based sentiment index using a dictionary-based approach and apply it along with a rating score as two proxies for obtaining residents’ sentiment toward their dwellings. To analyze the association between residents’ sentiments (text-based sentiment and rating score) and housing price, we employ hedonic pricing models and machine learning algorithms. Results indicate that the rating score is not a sufficient statistic, and combined use of text-based sentiment and rating score can explain housing prices more effectively. Methodologically, machine learning algorithms (i.e., random forest and extreme gradient boosting) exhibit superior performance over traditional hedonic pricing models in terms of accuracy. We also calculate SHapley Additive exPlanation values to identify the relative importance of each variable and confirm that text-based sentiment and rating score are significant when controlled with hedonic variables. Policymakers can employ the proposed text-based sentiment framework as an additional tool to monitor residents’ sentiment and adjust interventions in response to substantial sentiment changes in the market, enabling more targeted and adaptive housing policies. Additionally, incorporating residents’ sentiment provides traders with valuable complementary insights, enhancing the accuracy of valuation models. By utilizing both text-based sentiment indices and rating scores, which significantly influence housing prices while reflecting different aspects of residents’ sentiment, market analyses become more comprehensive and reliable.</p>