PRE-PROCESSED LATENT SEMANTIC ANALYSIS FOR AUTOMATIC ESSAY GRADING

Authors

  • Ruth Ema Febrita Magister of Computer Science Brawijaya University, Indonesia
  • Wayan Firdaus Mahmudy Magister of Computer Science Brawijaya University, Indonesia

DOI:

https://doi.org/10.28961/kursor.v8i4.110

Keywords:

Automatic Grading, Essay Grading, Latent Semantic Analysis, preprocessed LSA

Abstract

In education, essay is considered as the best tool to evaluate student’s high order thinking and understanding. In the other hand, manual processing and grading essay answers by a teacher need much time and tending to subjectivity grading. Meanwhile automatic essay grading in e-learning system find the difficulties in comparing model or key answer to student’s answer because student’s can answer the question with so various way. That means a right answer also can be so various, for they have same semantic meaning. This paper proposed automatic essay grading using Latent Semantic Analysis. But before the texts being scored, they will be pre-processed using stop words removal and synonyms checking. Calibration process implemented for dealing with the various possible right answer and help to simplify the term matrix. Implementation of this approach using Java Programming Language and WordNet as lexical database for searching the synonyms of every given words. The accuracy obtained by this method is 54.9289%.

Downloads

Download data is not yet available.

Downloads

Published

2016-12-20

Issue

Section

Articles

Citation Check