With the rising of intelligent agriculture and big data intelligence, the intelligent recommendation method of e-commerce platform for agricultural products has becoming an important measure to satisfy the personalized needs efficiently. Aiming at the problems of long time-consuming and low efficiency of traditional recommendation methods, this paper proposes an intelligent recommendation method of agricultural products based on big data processing technology. In this method, a kind of LDA-MF hybrid algorithm was formed by integrating the document theme algorithm and matrix factorization algorithm. Second, weighting the fusion of collaborative filtering algorithm based on the item and LDA-MF hybrid algorithm. Finally, a Spark parallel computing platform is built to capture the sales scoring and commentary data of jd.com mall and China Agricultural Products Network, then feature extraction, weighted fusion, intelligent recommendation and error evaluation will be carried out. Experimental results show that the LDA-MF hybrid algorithm can effectively improve the precision of recommendation, the theme weighted fusion collaborative filtering algorithm can improve diversity, and the intelligent recommendation method of agricultural products has been obviously improved in recommending quality and execution efficiency.