Abstract
To improve the veracity, several teachers are usually required to evaluate and mark the same summary in an examination. However, it will exhaust much time and energy and therefore exert a heavy burden on them. If a reliable auto−evaluation system is available to assist the teachers, on the one hand, it will reduce their work greatly；on the other hand, it will opportunely and rightly give the mark to learner without teacher's help, so the summary could be corrected and marked repeatedly by learner. Thereby, this system will enhance the learner's summary ability. The purpose of this paper is to discuss the feasibility of the system's applications in summary auto−evaluation with a view to Latent Semantic Analysis（LSA）and validate its usability. In this paper a model is established to elaborate how to conduct the summary auto−evaluation by taking advantage of LSA, the analysis result of this model shows the usability in contrast with the method of Tsubakimoto and Yoshimoto.
