Home page recommendation algorithm based on user eating behavior modeling

Abstract

The invention discloses a home page recommendation algorithm based on user eating behavior modeling. The method comprises: step one, all application operation behavior data of a user are obtained by means of application terminal point burying; step two, a clustering analysis is carried out on all user data by using an SVM to form a pre classification unit; step three, it is assumed that user behaviors in each classification in the pre classification unit are similar, modeling is carried out on users of each classification, all user data in the classification are learned by using a neural network for a user behavior model in each classification, and repeated iteration is carried out to obtain an optimal solution; and step four, latest behavior data of the users are substituted into the obtained models and recommendation information of the users is obtained in real time. According to the invention, difficulties of relative solidification and difficulty improvement of recommendation conversion rate in the prior art can be solved. On the basis of SVM pre classification, a classification user model is determined and dynamic recommendation is carried out based on the real-time algorithm model; and the maintenance cost is lowered and the home page recommendation conversion rate is improved.

Description

technical field [0001] The invention relates to a personalized recommendation technology, in particular to a home page recommendation algorithm based on user dining behavior modeling. Background technique [0002] At present, various types of dining platforms and applications are emerging in an endless stream. How to better proactively push suitable restaurants to target users and induce them to order meals is an important proposition for catering-related applications. [0003] The following is an example of a food ordering application. Usually, on the homepage of the application, there will be a special section for displaying restaurants recommended to users. Commonly used methods are mainly cut from the following aspects: [0004] 1. Considering the profit margin of the company itself, the restaurant that the company itself can obtain high profits will be recommended on the home page; [0005] 2. In consideration of the hotspots of user consumption behavior, put the rest...

AI Technical Summary

Problems solved by technology

Even through the analysis of a large amount of data, there is still a high possibility that due to the large behavioral differences among users and the overfitting of the analysis model, the final recommendation will be completely different.

On the other hand, most of the above-mentioned recommendation methods need to be set through the background, and are only valid for a certain period of time, requiring continuous updating by manpower, which loses timeliness

The conversion rate of end users by recommending restaurants cannot be significantly improved, and manpower is required to continuously maintain the corresponding recommendation functions

It is difficult to provide real value to users, and it is even more difficult to increase user conversion rate and revenue for enterprises

Images