There are two main components to leading the discussion of a topic in class: The presentation, and managing the discussion.

The discussion leaders should make a brief presentation summarizing the main points of the papers. You may assume that everyone read the papers. It is not your job to help students who came to class unprepared. However, you should not assume that everyone understood all parts of the papers, or that they understood what you consider to be the significant points. Do not bore your audience by stating things that everyone should know from reading the papers. Instead, go straight to the key points, and go over them quickly, so that any misunderstandings can be addressed before the main discussion begins.

TREC papers are frequently rushed and incomplete, so a paper may explain some technique only briefly or unclearly. Usually the research reported in the paper is not an isolated project. Instead, it is part of a project or PhD degree that spans several years. If something is unclear, check the authors' prior publication(s) for clues about what was done. As the discussion leader, this is your responsibility. You do not need to conduct an exhaustive literature review, but do check what they have published in the previous year or two.

Try to keep your presentation of both papers to a maximum of about 2/3 of the class. Everyone will be happier if most of the time is spent discussing (arguing about) the papers.

Being a good discussion leader is a mix of skills. Some are summarized below.

1. Arrive in class with about 10 topics that are worth discussing (see below). Prioritize them. Probably the class won't get to them all, but ... you need to be prepared with an interesting topic when discussion lags.
2. A good discussion typically focuses on important or controversial issues.
3. It does no good to lay out 10 issues at the beginning of class. People will forget what you said. Pick one, and let the discussion focus on it for 5-10 minutes. If the discussion begins to lag, or if your internal time limit is reached, wrap up the discussion of that issue by summarizing the key points and positions, and then move on to the next issue.
4. Try to keep the discussion focused, but if it naturally wanders off onto an appropriate topic, that is fine, especially if it is a topic that was on your list of topics to discuss.
5. Try to not to let one person or a small group dominate the discussion. It is fine for a small group to dominate the discussion of some particular issue - they may have the most interest in or knowledge about that topic. However, the rest of the class will become bored if this goes on too long, so try to make sure that a different group is involved when you shift to the next topic.
6. Managing the discussion does not mean that you need to respond to every point that anyone makes. You are not the focus of the discussion. Let others talk. Let others respond to questions or comments. If others are doing a good job of talking, you don't need to talk. If the discussion is lagging, or if it is time to change the direction of the discussion, then you need to be involved.
7. Sometimes a discussion will become overly impressed or (more likely) overly unimpressed with a paper. It is not unusual for students to begin picking apart a paper, and to decide that it is trash. Remember that these papers were picked because we think that they represent the best work done that year. Try to strike a balance between what was done well, and what could be improved in followup work.

To summarize, managing a discussion is fairly easy. Give people something interesting to talk about, and then get out of the way!

Copyright 2011, Jamie Callan.
Updated on December 10, 2011.