A Novel Feature Clustering Algorithm for Evaluation of Descriptive Type Examination

Objective type of Examination evaluation is easy in Computer world, but the descriptive type of question evaluation is more complex and there is no significant research has been taken place. So many descriptive type examinations like University Exams, GRE etc. , have been conducting from long time which is being evaluated manually by sending these types of questions and answers to the experts. This kind of exams needs automatic evaluation instead of manual correction to bring accuracy and reduce the evaluation time. In this paper, authors propose CosInfo algorithm a new solution to the above problem which can evaluate the papers automatically. This algorithm implemented the feature clustering for evaluation purpose that calculate the similarity between two documents and cluster the relevant documents in to different groups. Proposed algorithm uses the expected information function and parts of speech in English grammar as parameters to cluster the data, and also builds a model to classify the testing documents using SVM classification to assess the degree of similarity which will help to award the marks automatically. Experimental results show that the proposed method obtains better and accurate results to allocate marks compared with manual evaluation.