Construction of Machine Learning Model Based on Text Mining and Ranking of Meituan Merchants

Abstract

In the Web 2.0 era, the problem of uneven quality and overload of online reviews is very serious, and the cognitive cost of obtaining valuable content from them is getting higher and higher. This paper explores an effective solution to address comment overload by means of information recommendation in order to improve the utilization of online information and information service quality. This paper proposes a review ranking recommendation scheme that focuses on the information quality of reviews and places more emphasis on satisfying users’ personal information need. The paper’s approach is used to extract and rank low-frequency keywords that appear only once in the comment set. The more useful the extracted phrases are, the more useful this review will be and the higher the usefulness votes will be, which can reflect the actual situation of this product more objectively and accurately and facilitate better consumption decisions for consumers. The experimental results show that users’ satisfaction with the perceived usefulness of the reviews is jointly influenced by the information quality of Meituan’s reviews and users’ individual information needs; the recommendation strategy achieves the organic integration of the two, and the evaluation results under three different recommendation modes show that compared with “interest recommendation” and “utility recommendation,” the satisfaction score of “fusion recommendation” is the highest