A digital pen-based writing state recognition algorithm for student performance assessment

Technology-enhanced learning is an irresistible trend in intelligent education. However, most digital pen-based studies focus on handwriting character recognition, writing behavior research are extremely scarce. In this work, we prototype an embedded digital pen aimed at classifying students’ writing behaviors. Utilizing state recognition, feature extraction and optimized k-means modeling method, we present a WSR (Writing State Recognition) algorithm. WSR can classify writing and short-writing indexes. One hundred and eighteen juniors participated in the algorithm validation. Experiment results show that writing behaviors are strongly correlated with the test scores. Our proposed WSR algorithm can help teachers grasp students’ writing status, assess performance and acquaint learning emotions. The digital pen-based assistant application can shed light on personalized teaching and also has great prospects in the future education.