Content Extraction Model of Bilingual Loose-leaf Textbooks for Vocational Minority Students Based on Neural Networks

The semantic feature vectors of teaching texts were determined for information extraction using the characteristics of bilingual loose-leaf textbooks for ethnic minorities in vocational colleges. The corpus of bilingual loose-leaf teaching was trained in neural networks to extract template semantic feature vectors independently. Then, the required data was extracted in task modules. In this method, lexical analysis was not necessary owing to its relatively fine information processing and granularity. The algorithm proposed in this study showed an improved accuracy upto 30.5% compared to those of traditional algorithms. It also had the adaptability to standardize language semantic elements and structures.