Automated Essay Scoring menggunakan Cosine Similarity pada Penilaian Esai Multi Soal

Abstract

Process of determining scores automatically from one or several documents based on text data included in the field of Automated Essay Scoring (AES). Mechanism for scoring answers manually takes a long time to assess students' essay answers, especially for multi-question answers. The implementation of Automated Essay Scoring (AES) is to make correction and scoring easier by application that run on computers. This study applied TF-IDF weighting method and cosine similarity measurement method in multi-question of essay answer document. The test carried out through the preprocessing phase for data extraction, then weighting on each word term. Weighted value is the calculated using cosine similarity to get the degree of similarity. In each question, the point has been given, the cosine similarity value can be calculated with points from each question. This process produces a final score from student answer document which is compared with an expert document. By using the principle of similarity, only the relevant text or character from an expert document will be given weight, to get the similarity value, the number of terms contained in the answer document must match, neither in fewer nor more in number.