Data Mining and Exploration

Paper Presentations

The second part of this course includes paper presentations from student groups of 3. Every group will be allocated a paper of their choice and should prepare a presentation on it.

Here is a list of suggested papers for the presentations. You can also choose a paper outside of this list, but you must clear it with the instructor by email.

Group Allocations

Group No.	Members	Paper	Presentation Date
1	Angus Scott Debadri Mukherjee Zhou Yu	Connecting the Dots Between News Articles [pdf]	Feb 27
2	Imran Khaliq Rajeev Ratan Aart Meijer	Meme-tracking and the Dynamics of the News Cycle [pdf]	Feb 27
3	Grant Robertson Anthony Jarvis Dan Benveniste Mustafa m Somalya	A Hierarchical Bayesian Markovian Model for Motifs in Biopolymer Sequences [pdf]	Mar 6
4	Junjie Ye Jin Li Guannan Lu	Group Formation in Large Social Networks: Membership, Growth, and Evolution [pdf]	Mar 6
5	Daniel Zapata Jesus Em. Vazquez Valencia Jose Sendra Cesar Juarez Ramirez	Object Recognition with Informative Features and Linear Classification [pdf]	Mar 13
6	Shashank Mangla Iain McDermid Angus Taylor Matthew Gould	The Author-Topic Model for Authors and Documents [pdf]	Mar 13
7	Qiuyu Li Qi Hu Ziwei Peng	Suggesting Friends Using the Implicit Social Graph [pdf]	Mar 20
8	Xiaohong Zhao Shufei Zhang Yunfeng Zhu	Modeling Relationships at Multiple Scales to Improve Accuracy of Large Recommender Systems [pdf]	Mar 20

Guidance / Instructions

The following guide was written by Amos Storkey. It should be very helpful for preparing your presentations.

I would also suggest reading the instruction found here.

General Rules for Presentations

A presentation should have a point. Think of it like an elevator pitch. What is the one sentence conclusion you want the listener to take home at the end of it all? If you don't have a one sentence summary in your mind, then your presentation is doomed from the start. Note you don't have to say your one sentence summary. But you do have design your talk around it. Presentations are potentially different from lectures in this way, as lectures need to relate to a syllabus. Even then good lectures try to work to a point.

The need to establish a point is the most important thing about presentations, and yet very few presenters are ever explicitly taught this and many fail to do it. Remember this one for the rest of your life. It could be the most important thing you learnt from the MSc :).
The rest of the presentation should relate to the point. Each think you say should elaborate the meaning of the point, provide argument to establish the point, provide examples to illustrate the point. If it doesn't relate to the point it shouldn't be in the presentation. Showing people things is more important than saying things.
A presentation should not cover everything. It is an advert for the paper (though not the sort of advert invented by marketing people!). It should convince the reader that they want to know more, and give the reader enough to know what the paper is doing and how it is doing it.
The presentation should tell a story. It has an introduction, a middle, a climax, and ending. Much of what you were taught about creative writing holds for presentations too. This is hard though. Very few presentations are as riveting as a good novel :).
Giving a presentation is not the same as reading from a script. However for those for whom English is hard, it is quite understandable that you might wish to script what you say. If this is the case try to say what you want to say first, and then script it the way you would say it. A good way is to record giving the presentation without a script and then write down what you chose to say, but there are other ways. When presenting, don't be afraid to deviate from a script. The more you can escape your script, (unless you are a trained actor) the better off you are. If you simply remember the main point for each item you have to say, then so long as you get that main item across, you have done the job. It matters less what you say, more what you show.
Slides should support a presentation. They should not be a presentation. A presentation is your communication with the audience, not your set of slides. Pictures and diagrams are often more supportive than words (you can say words, you can't say diagrams). Don't overdo the time on presentation slides at the expense of knowing what you are going to say. "Less is more" on slides.
Presentations are hard to give. Giving top presentations takes vast preparation and practice. Even those who can give v. good presentations often don't if time is a limitation. Don't expect too much of yourself.
Presentations are hard to listen to. Most people find talks soporific. They are tempted to fall asleep even if interested... Anything you can do to keep people awake is good!

Specific ideas for DME

Work together. Decide on the point. Carve the paper up into different bits, get people to present self contained parts.
OR Make the paper a discussion between people. Have a running theme with the different people taking different parts.
A presentation need not match the order of the paper. Its best to choose your own order using the advice above. Sometimes this turns out to be the order of a paper (which isn't too surprising). However talks are more flexible. In a paper, if X is dependent on Y it must follow it. E.g. examples follow theory. In a talk you could give the examples first e.g. if you think the precise applications and results are more important motivation than the method. (i.e. Look what this approach can do! This is how they did it...)
Context matter. What is your point? Is it really about the specific paper? Or is it more generally about these types of methods in a field, with this paper an example? Decide on your point. Stick with it.
Relax. If you put the work in, so you know what you are going to say, then its likely to go fine.
Allow time for questions, and find ways to encourage questions.
There is no 7.
You may want feedback on your presentation from the audience to help you in future presentations. If you want this then please state that at the beginning of the talk and allow some time at the end. If you don't what this, then that is fine too. If you are in the audience and are giving feedback then please be constructive - remember it is not easy. Negative statements don't help anyone. We are all here to learn.

This page was revised by Stefanos Angelidis and is maintained by Nigel Goddard

Home : Teaching : Courses : Dme : 2015