Every college, university, school conduct exams and most important part of exams are the results. In order to get these results, the exam papers have to be evaluated one by one manually. This process of evaluating the exam papers is time consuming and requires more manpower. To overcome this solution, we have come up with a thought that removes the manual evaluation process. Our project focuses on developing a system that evaluates an answer script against a pre-uploaded marking scheme. Initially, a model is trained with a huge corpus (text data) and this model is used for further comparison of answer sheets with marking scheme. For evaluation, the marking scheme is processed and created a bag of words. In the same way, even answer sheets are processed and created a bag of words. Now these 2 values are compared using the LSI model and marks are assigned to the answers according to the comparison percentage obtained.