抄録
Software development begins with writing a requirement definition document (RDD) specifying what software is to be built, and the RDD should define the necessary and sufficient conditions the software must satisfy. Preferably, the RDD is reviewed to guarantee its quality. However, the quality of reviews of such documents is not easy to evaluate, due to various review styles and the logical complexity of RDDs. Therefore, we regarded the review ability as the ability to match an RDD with a software and developed a game to make a matching task that can assess review quality. The task has four types of relationships, two-by-two classes of necessity (i.e., the RDD has no irrelevant sentences to define the given software) and sufficiency (i.e., the RDD covers all the parts of the software), between requirements and software, which are expressed in verbal and nonverbal forms. Results suggest that the game likely sufficiently simulated the process of making/reviewing RDD in the requirement definition process. Therefore, it is suggested that the matching task created through the game can be adequate to assess the review ability.
