Informatics Research Review

1
Informatics Research Review

• Alan Bundy
• University of Edinburgh

2
What is it?

• Survey of research in targeted area.
• May be forerunner to summer project.
• Supervised by project supervisor.
• Approximately 3000 words.

3
Why do it?

• Learn skills of research reading.
• Learn skills of research writing.
• Confirm choice of research area.
• Learn background to project area.
• Compulsory course in your MSc degree.

4
What to do

• Identify relevant papers and/or seminars.
• Read the papers in varying detail.
• Keep notes on each paper/seminar.
• Weave these into a story.
• Write your report.
• Submit by 3rd December 2004.

5
Structure of Review

• Introduction identify and motivate topic.
• Main Body
• Summarise each piece of work.
• Give critical analysis.
• Compare and contrast.
• Conclusion
• What is state of the field?
• Where next?
• Bibliography list all (and only) papers cited.

6
Relation to Project

• This might form literature survey of project.
• But choice of project still open.
• Allocated supervisor reflects your interests.
• But may not be proposer of your preferred
  project.

7
How to Identify Papers to Read

• Project proposals will cite papers.
• Follow-up the citations in these papers.
• See who cited papers you read.
• Search for related work
• on the internet,
• in the library.
• There are library and online resources.
• Library Online www.lib.ed.ac.uk/resources.
• Citeseer and ISI Web of Knowledge.

8
Reading/Listening to a Purpose
Always have questions in mind when reading or
listening.

• What are the aims and objectives of the work?
• What was achieved?
• What claims are being made?
• Is the supporting evidence convincing?
• Where will my project fit in?

9
Reading to Different Depths

• Some work is central to your concerns, some less
  so.
• Need to vary reading depth.
• Some need only skim.
• Some read in depth.
• Some in between.
• Could be 20 papers in total,
• but only 3 or 4 in depth.

10
Typical Claims in Informatics
X is better than Y on task Z along some dimension
W.

• What kind of things are X and Y?
• system, technique, parameter?
• What is task Z?
• What is the dimension W?
• behaviour, coverage, efficiency, usability,
  dependability, maintainability?

11
Hypotheses in Informatics

• Hypotheses/claims often not stated.
• except in theoretical work.
• Absence leads to confusion and misunderstanding.
• If claim not clear then this should be
  criticised.
• Evidence may be theoretical or experimental.
• Objective may be to identify a hypothesis for
  subsequent evaluation.

12
How to Skim

• Read title, abstract, introduction, conclusion,
  bibliography, key sections.
• Identify main contribution of paper.
• How does it relate to other work?
• Identify key questions to be addressed and hunt
  for answers.

13
How to Read in Depth

• Make several passes over the paper.
• Start by skimming.
• Then read in increasing detail.
• Apply techniques to your own examples.
• Try explaining the ideas to a friend.
• Try the ERA referee assistant.
• www.inf.ed.ac.uk/teaching/courses/irm/reviews/era/
  

14
Seminars

• The School has several seminar series.
• Run by institutes, programmes or groups.
• Find list at www.inf.ed.ac.uk/events/seminars/.
• Supervisor can advise on which to attend.
• Infeasible to base report solely on seminars.
• Listen with question in mind.
• You can ask this question.

15
Why Keep Notes?

• Helps focus your ideas as you read/listen.
• Otherwise, you will forget and need to reread.
• Acts as zeroth draft of your review.

16
Telling a Story

• Literature survey is part of motivation.
• How did this field develop?
• How did it start?
• What are the rival approaches?
• How do pieces of work relate?
• Where are we now?
• What remains to be done?
• What are the hot topics?

17
Avoid Plagiarism

• Quotations must be acknowledged.
• Including close paraphrase.
• Use quote marks and cite source.
• Best practice in research.
• Read School guide on plagiarism.
• www.inf.ed.ac.uk/admin/ITO/DivisionalGuidelinesPla
  giarism.html.
• Plagiarism carries serious penalties.

18
Bibliography

• List all and only papers cited.
• Hacker 2000 Hacker, A., A model of free will,
  Journal of Computational Theology, pp 1-42,
  Vanity Press, 2000.
• There are several styles and publication types.
• Latex supports several.
• Provide the following
• author(s) name(s), paper title,
  journal/book/conference title, year, pagination,
  volume/number, editor(s), publisher.

19
Pacing Yourself

• Work out timetable for reading/writing.
• Leave plenty of time for feedback and correction.
• Read at a steady pace.
• Write as you go.
• Meet with your supervisor regularly.

20
Assessment

• Your report will be marked by your supervisor.
• Assessment will be based on
• Appropriate coverage of the topic.
• Demonstrated understanding of sources.
• Critical evaluation and comparison.
• Coherence of story.
• Clarity of expression and presentation.

21
Conclusion

• Make thoughtful and thorough search for sources.
• Papers and seminars.
• Study each in appropriate depth.
• Keep notes.
• Critically evaluate and compare.
• Weave into story.
• Write 3000 word report.
• Reflect story and relevance in report.
• Leave time for feedback and correction.