A heterogeneous opinion-driven decision-support model for tourists' selection with different travel needs in online reviews

The advancement of tourism websites has greatly improved the travelling experiences of tourists. One such way is the recommendation of restaurants by tourism websites. As a result, online restaurant reviews have grown tremendously in recent times. Using online reviews of restaurants in Ghana, this article deeply examines tourists' restaurant experiences. Specifically, we employ unsupervised machine learning techniques such as Term Frequency-Inverse Document Frequency (TF-IDF) and K-means algorithms to detect restaurant factors and evaluation attributes. We further develop an improved probabilistic linguistic linear programming technique for multidimensional analysis of preference (PL-LINMAP) to derive the attributes' weight importance for different tourist groups. Additionally, we propose an uncertain decision-support model known as probabilistic linguistic Measurement Alternatives and Ranking according to the COmpromise Solution (PL-MARCOS) to aid different tourist groups in satisfactory restaurant selection. This study provides a comprehensive model for restaurant managers in understanding heterogeneous tourist preferences.