Power Integration Mall Recommendation Model Based on Reverse Reward Feedback Learning Optimization

At present, the recommendation algorithm of power integration mall is often limited to users' personal preferences and historical choices, and it is difficult to capture the change information of users' behavior exchange trend, so it is easy to lead the recommendation model into local optimum. In order to solve the above problems, a method of power integral merchandise recommendation based on inverse reward learning optimization is proposed. The algorithm constructs a latent customer mining model to cluster users in multi-dimension, extracts the relevant feature information of different user groups, and implements the convertible merchandise recommendation for different user groups with three-dimensional rating recommendation algorithm. Subsequently, an optimal recommendation model based on inverse reward feedback learning is adopted to guide the next recommendation behavior according to the reward plastic function formed by historical exchange records. On the premise of satisfying users' individual exchange preferences, on the one hand, it avoids the limitation of the traditional recommendation algorithm which leads to the recommendation of goods, on the other hand, it enlarges the range of users' choice of exchange by the direction of recommendation formed by the overall users' exchange preferences. The experimental results show that the model can improve the success rate of customer converting commodities, and illustrate its feasibility and effectiveness.