KWA: an improved question matching algorithm based on keywords' weight analysis

Question and answer is an important link in distance learning. The best way is using a Web answer machine. Current methods employ keywords and frequency of a question to match questions stored in the question database. The major drawback of this approach is that the precision is not very high while the quantity of questions which include the same keywords subset and similar frequency increases. The paper proposes an improved algorithm that takes different keywords' weight into account. The experiments show an improvement to current question matching algorithms merely based on keywords and frequency.