Dynamic POIs recommendation method based on TS24

Abstract

The invention discloses a dynamic POIs (Point of Interest) recommendation method based on TS24, and belongs to the technical field of computer application. The method comprises the following steps: 1, building a dynamic POIs recommendation architecture based on 24 time periods; 2, converting the sign-in frequency of a user into a preference value of the user for the sign-in POIs of the user by using a TF-IDF technology, and constructing a sample set theta cur in a current time period; 3, establishing a SemiDAE geographical influence model on the Tau cur; 4, according to the similarity of the sign-in behaviors of the users in the similar time periods, establishing a T-SemiDAEPOIs recommendation model with the time influence on the Tau cur; and 5, according to the two steps of pre-training and parameter fine tuning, training the T-SemiDAEPOIs recommendation model on the Tau cur. According to the method, the dynamic POIs recommendation model based on the deep learning technology is built in a novel and more reasonable mode, geographical and time information of the location social network is mined and fused, and experimental results show that the applied technology can remarkably improve the precision and recall rate of personalized POIs recommendation.

Description

technical field [0001] The invention belongs to the technical field of computer applications and relates to a TS24-based dynamic POIs recommendation method. Background technique [0002] The present invention takes Location-Based Social Networks (LBSNs) as the main research object. In the past few years, some social networks have failed, such as Brightkite, Gowalla, and Jiepang, and the main reason for failure is that the quality of personalized Point-of-Interests (POIs) recommendation service is not enough to meet user needs [1] . Personalized POIs recommendation can greatly improve the quality of location-based social network services and benefit both users and POIs owners. For example, users can find POIs they might like. POIs providers may also use targeted advertising or promotions for specific users. These recommendations not only enhance the stickiness of users, but also increase the revenue of POIs providers [1] . To some extent, the quality of POIs recommendat...

AI Technical Summary

Problems solved by technology

In the past few years, some social networks have failed, such as Brightkite, Gowalla, and Jiepang, and the main reason for failure is that the quality of personalized Point-of-Interests (POIs) recommendation service is not enough to meet user needs [1]

Images