Conclusions and Future Work

---

This chapter presents intelligent agents facilitating distributed collaborative learning. It covers agent design issues and implementation details. In our research, we provide different support to users (including students and instructors). We have combined awareness information and advice, agent regulation, students’ self-regulation, and instructor regulation. The performances of these agents have been explored in various scenarios, both in asynchronous and synchronous collaborative environments, and positive feedback was received from students and instructors.

Several issues merit our further investigation:

• Determining what feedback is useful to the students is an empirical question. A fine-tuning of content of messages is a collaboration between the instructors and the system developers, taking into consideration the student’s opinion about the usefulness of the feedback. This is one of the issues in focus during the current user testing. In our design, agent feedback is tailorable by the instructor, depending on what he or she wants the agent to say to the students. The instructor writes the feedback in a text file, and the agent loads the content and presents it to the students.

• A more extensive testing was planned in conjunction with a large field trial in the DoCTA-NSS project in the Fall of 2002. In this scenario, students in two Grade 10 classes, one in Bergen and one in Oslo, collaborate on gene technology through FLE3. However, due to technical problems at the schools, the test was not possible. We are currently evaluating the agents in a university course with over 50 students. Data are being collected from server logs, questionnaires, and interviews. For the effectiveness of the facilitator agent, we plan to measure how helpful the awareness and statistical information is for the students (student opinion) and how effective the advice is in regulating the collaboration (e.g., look at actual changes in use of categories, see if participation level has increased, etc.). We will also be interested in how the students respond to the way in which the agent presents the awareness, statistical information, and advice.

• Our aim is to build a plug-in agent to support distributed collaborative learning. Therefore, we need to consider the reusability of the agent. How could we improve the reusability?

• For an agent to effectively regulate the distributed collaborative learning, it is crucial for it to understand the interactions between computers and students and between students and students. Although a few efforts have been made to study this topic (Barros & Verdejo, 1999, 2000; Mulenbruck & Hoppe, 1999) and some progress has been made, it still needs further investigation. For example, when the students are keeping “silent” in the collaboration, there is no way for the system to know whether the student is reflecting or doing something else.

• It seems that with the facilitator agent, the collaborative-learning process is well regulated. However, one can ask if it is good to have this regulation, or is it better to give the students more flexibility? We hope the results of our experiment will also help us answer some of these questions.

---