Abstract
The experiment information repository includes the information related with experiment projects. After the experiment information repository is implemented into a computer system and made to be accessible through the internet, the experiment information in the repository can be more easily used by engineers, compared with technical papers or experiment reports. This paper uses the classes and attributes of the data model to represent the organization of the experiment information of the experiment information repository, and presents some criteria to evaluate the usage of the classes and attributes for the experiment projects. The class usage ratio and the attribute usage ratio are useful to identify which part of the experiment information for each project is missing. The class family usage ratio helps to understand the information stored in the class including lower level classes. These evaluation criteria are applied into the NEES and KOCED repositories, and then some rules are proposed to improve the organization of the experiment information. An example of the rules is that the attribute of a class whose attribute usage ratio is low should be removed, or needs to be reorganized if the attribute includes at its lower level the class whose usage ratio is high. The usage evaluation criteria and the rules can be applied to develop other types of information repositories.