Multi-round Tag Recommendation Algorithm for Shopping Guide Robots
Yuyao Shi · 2018
With the rapid advancement of artificial intelligence technology, various "service robots" and "can talk and listen" commercial service robots have gradually entered our daily lives. How to use the multi-round conversation ability of shopping guide robots to search for users' intentions and promote the transaction of goods is one of the new retail formats. This paper constructs a multi-round tag recommendation algorithm framework for shopping robots and forms a product label mining and link prediction scheme. Combined with CTR estimation of user information, online enhanced tag recommendation is realized by means of reinforcement learning. Through the comparison of experimental data, the algorithm model has increased the click volume of the product by 6%. The shopping robot can gradually understand the user's purchase intention through multiple rounds of dialogue scenarios, and then achieve personalized recommendation, which ultimately leads to the transaction of the product.