The main purpose of this research is to evaluate and assist students through excellent computer teaching supplement, and furthermore to study the essential role that the evaluation and test have always played in evaluating the efficiency of learner in the field of computer based test and internet learning. And this study uses I.R.T., C.A.T., and web tech to provide the function of data base management, online test, online learning. So we hope that we can create a new mode to carry out computerized testing. By the way we bring out a complete set of personalized test and evaluating strategy, which include selection and practice of test items according to Bloom's category theory to understand the overall performance on test items of contestants and to provide the function of evaluating the procession and recognition dimension of personalized learning. This study is trying to compare the differences of C.T.I.P.E. online version 1 and 2. Besides item parameter, we use I.R.T. based on I.P.E. and E.C.A. (estimation of contestants ability). We also compare the influence of the differences of numbers of contestants on differences of E.P. with ANOVA in actual information. The result turns out that there indeed is decisive contribution of numbers of contestants to the stability of E.I.P.
