A novel Customer Service Recommendation Algorithm for Power Users

Abstract Aiming at the large amount of power user data and the fact that the collaborative filtering recommendation technology fails to consider the relationship between users and customer service staff, a k-means clustering and user portrait recommendation method is proposed. This method firstly uses clustering technology clustering the power users’ portrait label vectors to gather similar users together, and makes recommendations based on the cluster to which the users belong. Secondly, through calculating the user portrait label vector similarity between the attributes of the service feature vector to establish the connection between the user and the customer service, and improve the traditional grading forecast method, the user and the personnel of the service of similarity index is integrated into it. Finally, the personnel of the service will be recommended from the two aspects of service quality and service fits.