The evaluation of personalized learning features can perceive the features of learning behaviors intelligently, and provide direct, reliable decision support for promoting personalized learning resources (PLRs). The current research has an urgent need to overcome several problems of PLR recommendation: the recommended PLRs fall short of demand, the learning behaviors are not analyzed dynamically, and the learning intentions are not predicted well. To solve these problems, this paper explores the big data-assisted recommendation of PLRs and teaching decision support. The education of spoken and written languages was taken as an example during the exploration. The authors detailed the flow of the proposed algorithm, and proved its effectiveness through experiments.