As the agriculture information developing, agricultural websites have become an important channel for accessing information for agricultural users, cooperatives and agricultural companies. Combined with Chinese modern agricultural business e-commerce platform, content-based filtering recommendation technology is proposed, adopting four-tuple to construct user interest model, introducing forgetting factor to mining and update user preference, and generating recommendations depending on the similarity of product model and preference model. By the practical tests, the results show that Content-based filtering recommendation algorithm can effectively improve the purchase rate.