Online Programming Semantic Error Feedback using Dynamic Template Matching

Many of automated computer programming feedback is generated based on static template matching that need to be provided by the experts. This research focusing developing an online semantic error using dynamic models students’ correct answers submission. Currently, there a lack model due their complexity and varies in terms structure. To solve formulation templates, new front rear n-gram sequence as technique was developed provide students missing structure best-matched template. We have tested 60 student’s Java 3 different types questions all templates randomly chosen for each student. An expert assigned manually match answer with templates. The result shows 80% student were similarly expert. Based matched template, will given notifying possible next instruction can included get it achieved contribute automatically assist answering computational exercises.