Research And Implementation Of Recommendation System Based On Latent Correlation Rating Model

With the rapid development of the modern tourism industry,online hotel booking has become the main way for people to choose accommodations.Getting information of interest through the Internet has evolved into everyone's daily behavior.The targeted comments published by the majority of users cover the important attribute characteristics of hotels,merchant service information,consumers' emotional tendencies and other valuable content.It has become an important reference basis for other potential customers to make consumption decisions.When consumers buy goods and services,they usually refer to the opinions of other consumers to determine whether the goods are suitable for their needs.Therefore,how to quickly obtain valuable information from the massive amount of information on the network has become an urgent problem that needs to be solved.The recommendation algorithm and related development systems are the important technical means to solve this problem.At present,many mature recommendation algorithms have been applied in many fields.The item-based collaborative filtering algorithm is currently the most widely used algorithm in the field of e-commerce.But it ignores a large amount of user text evaluation information.These text descriptions represent the true evaluation of the user and are of great application value.In order to solve the shortcomings of traditional recommendation algorithms,there have been researches using LDA models for extracting topic features.However,it performs text processing under the assumption that topics are independent of each other,ignoring the latent correlation between topic features.For the above problems,a latent correlation rating model was proposed,which introduces the cosine similarity theory to measure the potential correlation between topics,and defines close relative nodes to calculate feature weights.Furthermore,the potential correlation analysis of each review and the overall rating data are used to determine the feature rating of the review.Based on the latent correlation rating model,a recommendation system has been developed in this paper to provide users with more accurate information recommendations.The main work of this article is as follows:(1)The examination background and meaning of this subject are introduced,and the current condition of related technologies and research of topic models and recommendation systems by researchers at home and abroad is analyzed and compared.The main research content and overall structure of this paper are introduced in brief.(2)Aiming at the fact that the traditional topic model ignores the potential relationships between topic features when extracting topics,a latent correlation rating model was proposed,which introduces the cosine similarity theory to measure the potential correlation between topics,and defines close relative nodes to calculate feature weights.Furthermore,the potential correlation analysis of each review and the overall rating data are used to determine the feature rating of the review,and the validity of the model in this paper is verified by simulation experiments.(3)With Layui front-end framework and SSM back-end framework,a hotel service recommendation system is implemented through programming and the functionality of system is tested.