Personalized App Recommendation Based on Hierarchical Embedding
Dong Liu, Wenjun Jiang · 2018
We propose a personalized APP recommendation method based on hierarchical embedding. By deeply analyzing the potential hierarchical structures in the data, we use machine learning algorithms to predict the APP of interest to users.The relationship between the existing layers, find out the influence degree of the features in the hierarchy on the forecast results, and assign different weights to different features to obtain a personalized APP. Through the user's input of keyword queries or user's past records, and the matching of the weighted features in the hierarchical association, so as to achieve the purpose of personalized recommendation, we use machine learning methods to analyze and deal with the hierarchical relationship between users and APP, through machine learning method builds a hierarchical embedded model. Eventually, we mine the hidden relationship in the original data set. The predicted value of AUC reaches approximately 0.82. The layered embedded model makes it easier for us to find a personalized APP that meets the user and enhances the accuracy of APP recommendation.