Research on Data Fusion and Big Data Analysis Methods in the Optimization of College Students' Innovation and Entrepreneurship Education Curriculum System

Abstract

Knowledge mapping provides an effective solution to the problem of how to effectively and automatically integrate multimodal curriculum resources, as well as to better organize and present knowledge. In this paper, for the problem of optimizing the curriculum system of innovation and entrepreneurship education for college students, a named entity recognition model based on CNN+BiLSTM-CRF and an entity alignment model with fine-tuned BERT word embeddings are proposed to integrate the multimodal curriculum knowledge and visualize and analyze it. Experiments show that the Precision, Recall, and F1 values of the CNN+BiLSTM-CRF model in this paper reach 96.90%, 96.49%, and 96.69%, respectively, which are all better than those of other comparative models, which fully demonstrates that the proposed model in this paper has a considerable generalization ability. Meanwhile, the entity alignment model BERT+softmax proposed in this paper outperforms other models on each dataset, indicating that its design is more reasonable and can better serve the entity alignment task. In addition, the hot content of innovation and entrepreneurship curriculum research is innovation and entrepreneurship education, entrepreneurial ability and innovation, which can be the main direction of curriculum system optimization.