Implementation of the Jaccard Similarity Algorithm on Answer Type Description

Abstract

This study aims to measure the similarity of the answers to the description by using alternative answers as reference answers provided by the lecturer with a view to overcoming the diversity of student answers. This research focuses more on Indonesian language questions and answers by combining the jaccard similarity algorithm and keyword similarity. The results obtained indicate that by adding alternative reference answers, it can increase the correlation value to 0.78% and reduce MAE to 0.55. Likewise, after combining the jaccard similarity algorithm and keyword similarity, the correlation value increased to 0.78% and MAE decreased to 0.49.