Semantic similarity based assessment of descriptive type answers

The knowledge assessment and competence is an important division in an education procedure. An automatic evaluation system would be used to make sure the confidentiality and neutrality of evaluation. In this paper, proposed an algorithm called Assessment algorithm that uses semantic similarity for evaluation of detailed type answers. It will eradicate the inconsistency in the manual assessment. Also, for preprocessing proposed an algorithm for pruning and stemming which is used to reduce the size of the descriptive type answers. Stemmed words are converted to vectors using the semantic method, Latent Semantic Analysis (LSA). The vectors obtained from the semantic method are clustered using the Self-organizing map. Cosine similarity is used to measure the similarity between two vectors. Based on the value returned by the similarity measure, marks will be awarded. The proposed system will be experienced with detailed type answers written by the student. This method resulted in good precision, improved dependability of results, reduced the time and effort taken by the staff. The time complexity of the proposed algorithm is O (n).