Word Expansion using Synonyms in Indonesian Short Essay Auto Scoring

Exams conducted in online learning to evaluate learning processes have many formats, including essay format. Essays are considered more proper to measure learning activity results. However, essays require longer to assess student answers and have consistency problems if the assessment is carried out by different teachers or done separately. This study investigates the influence of word expansion using synonyms in Indonesian thesaurus on short essay auto scoring. The first step, reference answers and student answer text data is preprocessed by case folding, stemming, stop word removal, tokenizing, and duplicate word removal. Second, Word expansion using synonyms in thesaurus is used to generate alternate words for reference answers. Third step, the scoring process by calculating similarity and matching words. The score from the similarity and matching results is then used to generate the final score. Performance evaluation shows that the Dice Coefficient similarity method achieved the highest correlation by a very good correlation, and the smallest MAE was achieved by the Cosine Coefficient similarity method.