Online learning environments are associated with issues of the overload of information and uncertain lines of learning, and the traditional single recommendation approaches find it difficult to accommodate information aspects of multi-dimensional learning specifics. The current research develops a hybrid optimization of hybrid recommendation model. Based on the multi-dimensional data synthesis of the learner attributes, resource attributes and interplay behavior, an adaptive weight collaborative filtering and content vectorization fusion algorithm is built on, and learning effect feedback mechanism is provided so as to implement dynamic optimization of the environment. Experiments on data over real learning platform indicate that over the single collaborative filtering method, the precision is enhanced by 18.3; over the traditional content recommendation method, the precision is enhanced by 22.7. In the meantime, the time spent by the learners on learning is growing by 12.5% and the rate at which the courses are completed is also growing by 8.2%. This approach disintegrates the steady fusion mode of the conventional suggestion, actualizes dynamic environment adjusting as a result of learning response, and offers successful technical support to the precise development of customized study systems in internet education websites.
Ye et al. (Thu,) studied this question.