Many students in higher education are required to complete graduate work while learning state-of-the-art technologies and theories in their laboratories. To obtain their objectives efficiently, although it is important that students accumulate, share, and apply "research know-how", these are generally tough tasks because most students graduate without conveying their know-how to juniors. Inspired by the concepts of Knowledge Management (KM) and Human-Agent Interaction (HAI), we previously proposed a system to support the research activity of students. It presents appropriate know-how based on two types of model: the first expresses relationships between tagged information, the second represents features and trends of information retrieved and evaluated by the students. In this paper, we conducted several experiments using the prototype of our developed system: (i) to evaluate the validity of the user model constructed for respective users (students) based on their usage history of the prototype system, and (ii) to discuss whether the user model adequately estimates users' trends in terms of know-how retrieval/evaluation.