Personalized Hotel Recommendation Algorithms Based on Online Reviews

Online users’ review information can be an objective reflection of product features and values. An increasing number of consumers rely on online review information to make purchase decisions. Therefore, it is worth investigating how to extract effective information from a large number of online reviews to improve the personalization and accuracy of the recommendation systems. Due to the ambiguity of expressions, there is a high degree of uncertainty in users’ online reviews. In this essay, we introduce the theory of Probabilistic Language Term Set (PLTS) to describe the meaning of different terms and solve the ambiguity in online review information and propose a hotel recommendation algorithm. First, we use Jieba, Boson sentiment dictionary and manual annotation methods to perform sentiment analysis on online review sentences. Second, we use the PLTS tool to describe and statistically process the review information and generate a rating matrix. Third, we use the maximum deviation method to calculate the weights of different hotel attributes. Finally, we use the PLTS-corrected cosine similarity formula to calculate the similarity between hotels and generate hotel recommendation rankings based on user history. We selected 10 hotels in Zhuhai City as case studies and compared the results with other recommendation algorithms to verify the effectiveness of the proposed algorithm.