Application of LCF Model Based on Deep Learning in Campus Network Platforms

In the information age, schools want to develop better, they must pay more attention to multiaspect's evaluations and opinions, and the campus network platform is a good medium. We can get teachers, students and societies for the school's more real evaluation. In order to better analyze the emotional polarity of various groups for some aspect of the school, we need to build a model to analyze the emotional polarity of each comment. However, at present, most of the widely used models focus on the context of the whole text, but rarely pay attention to the linguistic context of a particular sentence, i.e., the local context. In other words, few people have paid attention to how aspectual affective polarity is actually related to the local linguistic context before. However, the advantage of the LCF mechanism lies in its ability to classify the emotions in a particular sentence from multiple perspectives while focusing on the local linguistic context. It mainly uses contextual feature dynamic masking (CDM) and contextual feature dynamic weighting (CDW) layers to mask the words outside the local context range, while in the embedding layer of the model we use the sharing mechanism to obtain the global context and local context feature information. In this paper, we apply the LCF mechanism to the field of education by applying it to a wide range of campus web platforms to obtain valid evaluation information quickly and accurately.