Aiming at the problem of low interest and low accuracy of traditional recommendation algorithm, this paper proposes a new recommendation algorithm based on implicit model. Firstly, we collect the feature values of the implicit model data to obtain the user's personalized preference information, and then judge and classify the relevance values of different information according to the collected feature data and search keywords. On this basis, according to the results of judgment and classification processing, different levels of information are recommended and sorted, and the steps of model information recommendation are optimized to realize the recommendation of implicit model information. The experimental results show that the user interest value of the recommendation algorithm based on the implicit model is higher than other traditional recommendation algorithms, and the accuracy of information recommendation is higher.