Using LSA for Automatically Assessing Free Texts
To evaluate the content of free texts is a challenging task for human. Latent Semantic Analysis (LSA) can be used to automate this process. The main idea behind the LSA technology is to extract the close relationship between the meaning of a text and the words that are present in that text. ShirWin Knowledge and Learning Systems Inc. is currently considering two instances of “free texts” assessment problems, one is automatic essay grading (where student written essays will be graded automatically given course materials and a set of human graded essays as training data) and the other is fact verification in a virtual learning environment (where a student plays the role of a occupational therapist needing to discover necessary facts from the patient file (or given information) in order to declare a diagnosis decision). Both problems involve understanding the inner meaning of the free texts which can be successfully accomplished by using the LSA technique. The main objective of the proposed project is to verify the effectiveness of the LSA technology and conduct extensive experiments to provide solutions for the two problems. The performance of the LSA models will be evaluated by measuring their correlation with the human-graded essays where better correlation will denote better effectiveness of the LSA technology.