Chinese composition automatic scoring method and teaching aid system

Abstract

The invention provides a Chinese composition automatic scoring method and a teaching aid system. The method comprises: a to-be-scored composition obtaining step; a shallow feature extraction step forextracting shallow features of the to-be-scored composition; a deep semantic feature extraction step for extracting deep semantic features of the to-be-scored composition, including wrongly written character features and grammatical error features; a scoring step for combining the extracted shallow features and deep semantic features, and adopting random forest fitting to obtain a scoring result of the to-be-scored composition; and a pinyin conversion step and a theme extraction step. According to the method, shallow features and deep semantic features of the composition are combined, high scoring accuracy is achieved, an ideal evaluation result is obtained through training on small samples, and the utilization rate of the samples is effectively increased; meanwhile, the functions of wrongly written character recognition and correction, pinyin recognition and conversion, grammar error recognition and correction and the like are added, multi-dimensional information is provided as feedback guidance of user writing, and user experience is enhanced.

Description

technical field [0001] The invention relates to natural language processing technology in the field of artificial intelligence, in particular to an automatic scoring method for Chinese composition and a teaching assistant system. Background technique [0002] Introduction to Automatic Essay Grading System [0003] AES (Automated Essay Scoring) is an educational auxiliary tool based on intelligent algorithms under the trend of the rise of artificial intelligence and deep learning technology. Compared with the manual scoring system, the composition automatic scoring system has the advantages of being more objective, timely, efficient and low-cost, and has received more and more attention and research. Therefore, the development of an automatic composition scoring system has gradually become a trend. The traditional automatic composition scoring system mainly uses shallow features to model and analyze the text, ignoring the deep semantic features of the text, while the deep le...

AI Technical Summary

Problems solved by technology

[0015] The deep learning technology represented by non-patent literature 1 relies on large-scale samples during training, and cannot achieve satisfactory training results in small-scale samples; the machine learning technology represented by non-patent literature 2 does not fully extract the deep At the same time, the fitting ability of the multiple linear regression model is limited, so the accuracy of composition scoring is low; the methods represented by patent literature 3 and patent literature 4 are based on neural networks, and patent literature 3 predicts by designing multiple neural networks Composition

Images