Indonesian essay grading module using Natural Language Processing

Moodle is a LMS application which serves as an online courses platform. As for now, Moodle doesn't have capability to grade essay question automatically, so teachers should grade all the submitted answers by students one by one. This is a problem for teachers because when two sentences are being compared, they don't only compare them syntactically, but also semantically. Our research proposes an essay grading module to help teachers in term of grading answers syntactically using common tasks in natural language processing such as formalization, sentence detection, stemming, POS (Part of Speech) tagging, and tokenizer. The developed module consists of one grader component and two plugins integrated to Moodle, one is “essayinagrader” plugin for the question type interface and the other is “igsuggestion” plugin for the quiz report interface. To grade an answer, “igsuggestion” plugin will use the developed grader component. This component utilizes INANLP, a NLP tool for Indonesian language, to process the chosen NLP tasks. In addition of INANLP, this component also uses WordNet thesaurus to compare synonyms and Jaro-Winkler algorithm to calculate string distance. Test results show that the developed module can grade answers with the same representation correctly. However, for answers with different representation, it will classify those differences as a false answer.