Survey of Techniques Used for Answer Evaluation using Semantic Network

Abstract

Abstract-Automatic Evaluation of Free Text Answers' have become a necessity, not only for better acceptance of online learning, but also to handle the pressure of assessment of a large number of students' responses in a fatigue free pedagogically correct method in traditional learning environments. This work is aimed at developing a model to evaluate free text answers of students based on the semantic similarity it has with the model answers prepared by teachers. The model answers are prepared prior to the evaluation process and through a process of dynamic semantic network building, a model is prepared which is used in evaluation. The proposed technique should allow the flexibility of comparing a student's answer with two or more model answers and finally evaluating it against the model answer it most closely resembles.