Contemporary research endeavors have evinced a substantial interest in integrating heterogeneous data sources within unified recommendation system frameworks. Concomitantly, the conventional two-dimensional product-user rating matrix ubiquitous in matrix factorization problems is being augmented by incorporating ancillary dimensions such as sentiment, temporality, and spatial characteristics. Concurrently, the challenge of surmounting limitations in capturing Vietnamese sentiment characteristics for data enrichment has garnered scholarly attention. Stemming from these two salient issues, the authors propound a hybrid model that amalgamates the factor matrix principle from collaborative filtering methodologies with sentiment analysis for prognosticating user rating propensities. Through empirical evaluation on a corpus of mobile application reviews, the proposed model has demonstrated its suitability for research purposes and exhibited superior predictive accuracy compared to simpler paradigms.Contemporary research endeavors have evinced a substantial interest in integrating heterogeneous data sources within unified recommendation system frameworks. Concomitantly, the conventional two-dimensional product-user rating matrix ubiquitous in matrix factorization problems is being augmented by incorporating ancillary dimensions such as sentiment, temporality, and spatial characteristics. Concurrently, the challenge of surmounting limitations in capturing Vietnamese sentiment characteristics for data enrichment has garnered scholarly attention. Stemming from these two salient issues, the authors propound a hybrid model that amalgamates the factor matrix principle from collaborative filtering methodologies with sentiment analysis for prognosticating user rating propensities. Through empirical evaluation on a corpus of mobile application reviews, the proposed model has demonstrated its suitability for research purposes and exhibited superior predictive accuracy compared to simpler paradigms.