Research On Performance Evaluation Of Institutional Repository With Mrakovreward Model

Institutional repository is based on the concept of open access,managing and publishing digital resources created by the school and its members,where teachers and students can get academic resources and academic exchanges. For universities, IR is feasible, highly cost effective and strategic way to establish cooperation with academic staff and promote academic exchanges. Currently,many domestic colleges and universities have established institutional repository, but the development is slow, the quality of service is not high. Thus, through technical improvements and more scientific management methods to improve its service level is necessary, these improvements should be established on the basis of scientific evaluation of performance. Simultaneously,by evaluating institutional repository performance level,we can find it defects in the operation and management, so we can improve it from the beginning and during use.By analyzing the characteristics of Markov reward models, proposed multi-state Markov reward models for performance measures of the institution repository in Chinese universities.We assume IRs state transition probabilities satisfy the exponential distribution,construction performance evaluation system based on availability,we get the transition probability matrix after determining state transition relations. Then according to the current state and transition probability matrix,we can draw the average availability Ai(t),mean number N/t) of IR failures during time interval [0, t] and probability of IR failure during time interval [0, t] Qi(t), thus evaluate the performance of institution repository in China.Take Xiamen University institutional repository as an example, by summarizing influence factors of performance level,proposes the evaluation target system suit, using Markov reward models to evaluate the performance level of Xiamen University institutional repository. Then we compared the results of the evaluation findings with other outcomes,we found that the results were very consistent. It shows that the model can scientifically measure the performance level of institutional repository and able to provide theoretical guidance to improve the quality of institutional repository.