An Automated Evaluation Method of Conceptual Data Models Considering Similarities of Attribute Sets

In database design, conceptual data modeling (CDM) is an indispensable activity. In CDM education, there is a problem that it takes much effort of teachers to evaluate learners' models. To address the problem, automated evaluation methods and systems for the models have been reported. Generally, they take an expected answer and the learner's model as inputs, and determine whether the model includes the necessary elements in the answer or not. However, they only consider the names of the elements to determine corresponding pairs of the elements in the answer and the learner's model. It is insufficient to evaluate the models in the similar manner to the teachers. Therefore, we propose an automated evaluation method of conceptual data models considering the similarities of attribute sets of the concepts. We conducted an experiment to compare the results of evaluation by instructors, our method, and a previous method. As a result, we found that our method is effective and valid to evaluate the models made by novice learners.