This paper proposes a paradigm of integration and optimization of English teaching resources based on the association rule algorithm and improves the Apriori algorithm by introducing interest measure and manual labeling through semisupervised learning of the neural network to improve the quality of English instruction assisted by the multimedia network. The efficiency of the method is higher than the original Apriori algorithm and the Apriori algorithm based on hash technology, according to experimental results. The new integration and optimization of the algorithm-based teaching model of English teaching resources also guide multimedia and network-assisted English teaching activities.