An Agricultural Information Recommendation Method Based on Matrix Decomposition Knowledge Graph Algorithm

With the promotion of the Internet of Things, big data, and other technologies in agricultural production, farmers are accustomed to searching for crop cultivation information through Internet. However, the traditional way of information search has problems such as time-consuming, inefficient, and inaccurate, so farmers urgently need an efficient, fast, and accurate information recommendation method. So this study proposes an agricultural information recommendation model based on the matrix decomposition knowledge graph algorithm (MDKG algorithm). It introduces the matrix decomposition algorithm based on the knowledge graph, which can constructs and trains the interaction matrix between users and agricultural information. It also solves the decomposed interaction matrix, and obtains the correlation score between users and agricultural information through calculation. Experimental results show that the MDKG algorithm is better than the DKN and RippleNet algorithms in AUC and ACC of information flow clicks and recall and precision of cold start recommendations, which indicates that the algorithm can deeply mine user preference characteristics to improve click-through rates and perform better in cold start scenarios. It can better alleviate the negative impact of data sparseness, which improves the efficiency of farmers to obtain agricultural information.