Post on 27-Dec-2015
Scientific Communication CITS7200
Lecture 10Giving a Talk
• Good presentation skills are vital in many professions and contexts
• Research seminars, conference talks, project meetings, budget meetings, sales talks, funding applications, job interviews, meeting your future in-laws, …
Key points
• Know your audience– What do they know?
What do they want to know?
• Figure out what they will learn and why they want to learn it
• Be professional– Organise yourself, your materials,
and your time
What to say and how to say it
• A good talk can usually be split into five parts
1. Definition and motivation of the problem
2. General theory3. Details4. Conclusions5. Question time
Define and motivate the problem
Defineand motivatethe problem!
Define and motivate the problem
• First impressions count!• Be concrete, not abstract
– Be concrete in motivation and in details– Generalise in conclusions, if appropriate
• Motivate with applications, philosophy, history, predictions, dollars, …
• Cover background only where necessary for understanding or for motivation
• Set up the notation and terminology– Keep notation (especially) to a minimum– Keep examples as simple as possible
• Remind, but do not assume• Emphasise your contributions• Often a “road map” to the talk
helps to impose structure
The general theory
• Describe the methodology• Describe the key results• Explain the significance of the
results• Sketch the proof/evidence of the
results– Pictures/graphs help
The details• Present one key result which is
– Important– Non-trivial– Representative
• Now give details– But this is the part of the talk that is
least important, and should be first to be cut
– Refer them to the paper/notes, if necessary, or to later discussion
The conclusion
• Put your results back in context– Once again, motivate the problem– Make clear your contribution
• Discuss any limitations• Discuss some interesting open
problems• Conclude: thank the audience, and
invite questions and comments
Handling questions
• Always allow time for questions– There are always questions!
• Allow/encourage questions during the talk, if you are confident enough
• Try to anticipate likely questions• Try to learn from the questions
Visual and aural aids
• High tech– PowerPoint, LaTeX, HTML, PDF, others
• Medium tech– Overheads, slides
• Low tech– Pen and plastic, whiteboard
General points
• Use large font sizes• Write points, not sentences• Do not put too much on one “slide”• Make the medium invisible
High tech
• Use presentation software if possible– Forces you to prepare in advance– Helps you to organise your talk– Makes changing your talk easier– Looks professional and well-prepared
• But plan for equipment failure, or portability issues– Have a lower-tech back-up plan
Medium tech• Overheads are good as a
back-up mechanism• But printed overheads can look
boring– And a mixture can look odd
• Always use permanent pens• With slides, video or audio:
be in control of the situation yourself, and prepare
Low tech
• Recommended only for confident speakers
• Allow you to tune the presentation on-the-fly
• Allow you to build up complex pictures/equations/graphs on-the-fly in any order, and to control the audience’s attention
Pointers
• Stick pointers are easier to aim– But don’t bang on non-rigid screens!
• Laser pointers give you more range– But avoid these if your hands shake
Voice issues• Face the audience, not the projector
screen, nor the computer screen, nor the OHP, nor the desk, …– And don’t get between them!
• Talk to (multiple) individuals, or at least to the back of the room
• Vary the tone and level of your voice• Use “normal” language
Getting through to the audience• Use repetition
– Remember that with a talk, people cannot “refer back” to earlier comments
• Use examples– Don’t get bogged down in
abstractions and generalities– Never present the details of proofs– Feel free to say “more details are given
in the paper/notes”
Know your audience• Scientists (or business people,
or the general public)– Will be interested only in general results
• Computer scientists– Use notation carefully and focus on the
problem• Theoretical computer scientists
– Will want to know exactly what your contribution is
• Experts– Will want to know how your work beats
theirs!
Timing
• Never over-run your allotted time– Talks which are too short are always better
than talks which are too long
• Be aware of how long you have, and stick to it
• Design multiple exit points, or identify material that can be skipped– Cut out details first
• Make sure you’re aware of the time
Handling nerves
• Everyone gets nervous• Make your nerves work for you
– Be animated: show them you care!
• But be prepared too– Deep breathing– Know your introductory “speech”– Use written reminders if necessary– Bring water, so you can pause and
re-group when necessary