Abstract
This paper presents an algorithm for automated marking of natural language responses of literal comprehension questions. It highlights the crucial differences between the content and style of questions for assessing different levels of reading comprehension, and argues that the literal question type can be effectively handled by largely syntax and structural based algorithm. The efficient algorithm compares student answers with model answers along a token-based approach. The small semantic variations in student answers made the omission of corpus-based approaches more sensible. The algorithm was evaluated with real data obtained from a local secondary school, and the performance was found to be very promising.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
