An Improved Recommendation Method Based on Content Filtering and Collaborative Filtering

Abstract

With the popularization of the Internet and the prevalence of online marketing, e-commerce systems provide enterprises with unlimited display space and provide customers with more product choices, while its structure is becoming increasingly complex. The emergence and application of the network marketing recommendation system have greatly improved this series of problems. It can effectively retain customers, prevent customer loss, and increase the cross-selling volume of the e-commerce system. However, the current network marketing recommendation system is still immature in practical applications, and the problem of data sparseness is serious. The problem of user interest drift is not well dealt with, resulting in poor recommendation quality and poor real-time recommendation. Therefore, this paper proposes an online marketing recommendation algorithm based on the integration of content and collaborative filtering. First, content-based methods are used to discover users’ existing interests. After that, the mixed similarity model of content and behaviour is used to find the similar user group of the target user, predict the user’s interest in the feature words, and discover the user’s potential interest. Then, the user’s existing interest and potential interest are merged to obtain a user interest model that is both personalized and diverse. Finally, the similarity between the marketing content and the fusion model is calculated to form a set of user ratings combined with characteristics and then clustered through K-means to finally achieve recommendation. Experiments have proved that this method has good recommendation performance.