With the advent of technology, online courses have emerged as an effective form of education. Discussion forum is an integral part of online courses. In discussion forum, the instructor posts a question and the students post answer to the question. The students can also comment on the answers posted by other students. Due to this complex structure of discussion forum, currently the assessment of the student for discussion forum is mainly based on statistics such as number of answers and comments posted, number of likes received and so on. But these statistics do not measure whether the answers and comments posted by students are relevant to the discussion or not. So to address this problem we have developed a system which uses Word Mover’s Distance (WMD) similarity measure to determine the relevance of answers and comments posted by students with the context of discussion. The system uses machine learning approach to determine polarity of comments in context with the answer. Then it generates a score for the answers and the comments. Finally the accuracy of the system is measured by comparing the system generated score with the human evaluation score.