Thesis and Dissertation How To Version 1.0

1
Thesis and Dissertation How To Version 1.0

• Anthony Skjellum
• HPC Laboratory
• Department of Computer Science
• Mississippi State University
• tony_at_hpcl.cs.msstate.edu

2
Disclaimers

• This talk is offered as is by a professor who
  is trying to help his students
• It is designed specifically for MSU/CS, but it
  may work well for other degree programs elsewhere
  too
• There is no perfect strategy or approach to
  success with a thesis or dissertation, your brain
  must stay engaged
• Your advisor is always your first line of defense
• There is no substitute for hard work, results,
  and planning

3
Why is this How To Needed?

• Computer Science Theses and Dissertations are
  scientific works
• A decided process is needed to address steps
  involved
• Open-ended and closed-ended aspects are involved
• Many stakeholders are involved
• Students do this once or twice at most
• Professors see the same process repeatedly
• Students previous experience evidently
  insufficient
• Desirement for Uniformly Higher Quality Papers
• Less confused, delayed, and unhappy students ?

4
Outline of this presentation, I

• Some terminology
• Outlining
• Academic Citation (Boring but important)
• Taxonomy for the entire process
• The Hypothesis
• The Preproposal

5
Outline of this presentation, II

• Preparing the thesis/dissertation proposal
• The hard work in between
• Preparing the thesis or dissertation
• Presenting the thesis or dissertation
• Overall Strategy
• Summary
• Other Resources

6
Terminology, I

• Thesis MS thesis at MSU
• Dissertation PhD thesis at MSU
• Taxonomy architecture of process used here
• Guidelines Refers either to Departmental,
  College or University Requirements for
  Formatting, Structure
• Requirements Refers either to Departmental,
  College, or University Requirements for
  Acceptance Processes for Theses/Dissertations
• Other terminology defined as we go

7
Outlining

• Forming the skeletal backbone and structure of
  any talk, paper, or related process

8
Outlining (A Process)

• Outlining is an important part of all
  presentation
• Outlining is not well understood by students in
  general
• A strong outline drives the structure of the
  entire thesis or dissertation process
• Outline must be done early and updated
  periodically (iterative refinement)

9
Outlining (What is it?)

• A structured set of bullets/sentences, reflecting
  the entire skeleton of a paper or presentation
• An organization for the presentation of ideas
• A structure within preview/review of ideas can be
  organized
• A structure within which tasks and
  accomplishments can be indicated
• A stable intermediate form that can be reviewed
  and OKd as part of the thesis/dissertation
  process

10
Outlining (A few rules)

• Topic 1 Balance and depth
• A. There cannot be an A without a B
• B. Single subheadings should be merged
• C. The depth of subbullets should be meaningful,
  not overkill
• Topic 2 Full sentence outlining vs. not
• Using full sentences can really help outlines
  prove useful in forming a first draft.
• Othertimes not
• Topic 3 Organization
• Be purposeful in what is outlined
• Make sure you revise your outline over time
• Dont hesitate to look how others have organized
  their work.

11
Where to apply Outlining

• To the thesis preproposal (itself an early
  outline)
• To the thesis proposal
• To the thesis or dissertation itself
• To all the academic papers you write

12
Outlining (Where to Learn More)

• The Chicago Manual of Style
• Books on term paper development
• On the web
• English TA/Tutor for your department
• Departmental Seminars for your major

13
Academic Citation

• Giving due credit to sources verbal and written
  in oral and written presentations

14
Avoiding Verbal and Written Plagiarism

• You must cite your sources in writing
• You must cite your sources when you give a talk
  (you must verbally give credit where due)
• If an idea, concept, or result came from a paper,
  book, person, or web page, give credit for it
• It is far better to err on the side of too much
  referencing you advisor can help you remove
  excess references if need be ?
• A set of references at the end of your talk is
  useless in the real world

15
Taxonomy

• An integrated, but informal process and strategy
  for the creation of the four to five key
  deliverables

16
What are the key deliverables?

• A hypothesis
• An MS preproposal (not normally required outside
  of my research group)
• A PhD dissertation outline document (not required
  for a thesis)
• A proposal
• A proposal presentation/seminar
• A thesis or dissertation
• A thesis or dissertation presentation (defense)

17
How are the deliverables related?

• Hypothesis scopes the entire Work
• Initial documents
• Preproposal includes Hypothesis
• Dissertation Outline includes Hypothesis
• Proposals
• Thesis Proposal refines Preproposal
• Dissertation proposal refines Dissertation
  Outline
• Proposal Presentation abstracts Proposal
• Thesis/Dissertation refines Proposal
• Defense Seminar abstracts Thesis/Dissertation

18
Relating Documents
PROPOSAL
THESIS/DISSERTATION
90
I. Introduction II. Literature Review III.
Approach IV. Experimental Plan V.
Status/Plans VI. Conclusions
I. Introduction II. Literature Review III.
Approach/Artifacts/ Related
Deliverables IV. Results V. Analysis of
Results VI. Conclusions/Future Work
99
90
50?
25?
reuse
19
A radical idea?

• It is often be better to write the proposal
  seminar before the proposal, it is an outlining
  exercise
• The thesis proposal then refines the seminar
• Unlikely youd follow this order with the
  thesis/defense seminar !?!

20
The Hypothesis

• Your proposed advancement of science and/or
  engineering framed concisely - either provable or
  disprovable in the context of your work

21
Hypothesis Development

• Requires iteration, trial and error
• Should be posed so it can prove or disprove
  something nonobvious
• Should not be negative
• Should not have avoidable vagueness
• Involving your committee is a good idea after you
  and your advisor think youre there

22
Hypothesis Example 1

• Through the application of a cache-friendly data
  structure (to be defined in this work) instead of
  row/column-major storage, the the performance of
  the Strassen-Winograd algorithm can be shown
  faster than conventional multiplication, over a
  to be specified range of problem sizes and
  shapes, whereas for such cases it is now slower
  than the conventional algorithm.

23
Hypothesis 2

• At least 10 more performance can be obtained in
  collective communication algorithms, for large
  message size, and large concurrency, through the
  application of multilevel algorithms that adopt
  to the network topology

24
Hypothesis Example 3

• On average, algorithms A, B, and C, are equally
  fast (or slow) in finding solutions to problems
  of class/description X,Y,Z.

25
Hypothesis counterexamples

• It would be shown that, if some changes were to
  be made to the current program, benefits occur
• Implementing code X on platform Y will yield
  performance Z
• No implementations exist that are faster for
  algorithm Q on platform R

26
The MS Preproposal

• Your one-page writeup used to frame the research
  activities with the hypothesis early in the MS
  process

27
Contents of MS Preproposal

• The Hypothesis
• Why the problem is interesting/important
• A little positioning relative to other work
• Activities to be Performed
• Questions to be Answered
• Probable Contributions of the Work (if successful)

28
The Dissertation Outline

• A five page document to highlight the plan to
  your committee before you undertake the work
  leading up to the PhD disseration proposal

29
Contents of PhD Outline

• The hypothesis
• Background on the research area
• Why the problem area is new, why the hypothesis
  (if proven/disproven) is non-obvious/useful/import
  ant
• Statement of the approach being considered

30
Preparing the Proposal

• Take your preproposal hypothesis
• Work in your area and learn a bit about it
• Develop the complete lit. review
• Outline and then write the proposal
• Give chunks to your professor (pipeline) if
  possible
• Revise based on feedback and your own new
  learning

31
Presenting the Proposal

• Not required to be public
• Choose content over form
• Reveal these things to committee
• Hypothesis
• What youre sure about
• What youre not sure about
• What risks/uncertainties exist
• How much progress youve made
• Seek meaningful feedback by making these issues
  clear in the presentation
• Make sure there is no disconnect with the
  proposal itself

32
The hard work in between

• This is the part I cant help you with

33
Preparing the Thesis/Dissertation

• 50 or more of the writing is already done,
  because of the proposal
• Define a thesis outline, derived from proposal
  outline
• Now, make sure that results are the key new
  ingredient
• Stay with your thesis outline once its mature
• Write clearly, and make sure you relate back to
  the hypothesis regularly
• Go back and get more data if you need to, as a
  consequence of the maturation of the document
  (logical completeness and what not)

34
Understand the Style Guide

• The department and university have jointly
  formulated its stylistic requirements
• These include rules about
• Format
• Presentation
• Standards
• Not about
• Quality of results
• Grammar/spelling/truth/beauty

35
Design your thesis or dissertation

• Look at previous successful models
• Choose an outline structure that meets the
  advisors requirements, and represents your
  results
• Refine that structure as you need to
• Relate the gross structure to your overall outline

36
Presenting the thesis or dissertation

• Things to do/not do

37
Presenting Ideas

• Make sure the hypothesis is up front
• Dont spend most of the time on prior work
• Make sure your hypothesis is connected to the
  results that prove or disprove it
• Dont baby the audience
• First 25 of talk for everyone
• Middle 50 of the talk for most everyone
• Wow the committee with the last 25 results
• Dont read your viewgraphs

38
Summary, Some Advice, and Conclusions

• Were almost done, here are a few more things in
  the spirit of review and looking forward

39
Timeliness

• Understand all the dates associated with
  graduation
• Plan the entire study forward and refine that
  schedule each semester
• When planning for graduation semester, work
  backwards from University/department deadlines to
  understand when your documents/results must be in
  place

40
Completeness (My incomplete list of issues)

• Meet all requirements (submit thesis, submit
  softcopy, get signatures on signature pages)
• Make sure the program of study is valid at the
  beginning of the semester you plan graduation
• Checkup on your advisor to make sure all forms
  are fully executed and on time
• Make sure you apply for graduation
• Make sure you have no academic holds

41
Overall Strategy

• Understand how the deliverables and
  activities fit together and how to march alone
  toward a successful set of results ?
• Dont be afraid to change structure as you learn
  more about the outcomes of the actual research
• Dont over edit single paragraphs or pages when
  big results are still waiting to come in
• Write as you go, not all at the end

42
Summary

• Thesis/Dissertation is a process
• Outlining is a key technique in the process of
  framing ideas, organizing your work
• Planning (outlining activities in time) also
  crucial
• A set of activities yield specific deliverables
  (hypothesis, preproposal, proposal, thesis, and
  two seminars along the way)
• Completeness of all requirements and procedures
  is a major stumbling block for students

43
Parting Advice

• Look at other peoples successful deliverables
• Study their outlines/structures
• Ask them questions (if theyre still on campus)
• Review strong and weak examples with your advisor
• Attend thesis and dissertation proposal
  presentations (if public) and defenses
• Learn to ask (and regularly ask) questions at
  presentations
• Observe how the candidates handle questions
• Review/critique this presentation with your
  advisor
• Pay attention to deadlines and requirements

44
More Parting Advice

• There are three ways to solve a problem know the
  answer, know where to look, or whom to ask.
• Be prepared to say I do not know if you want to
  get help from your advisor or other knowledgeable
  professional (or even directions on a country
  road)
• Learn the implications of Occams Razor
• The Principle of Optimal Sloppiness Applies to
  All Human Endeavor (Max Delbrück, Caltech Nobel
  Laureate, 1936) If you are too sloppy, you
  never get anything done if you are too neat, you
  never get anything done there is an interior
  optimum. It is a corollary of Occams Razor.
• Scientific inquiry has dead ends. It is
  important to recognize dead ends, and move on
  quickly.

45
Other Resources

• Google is the best web search engine
  www.google.com (search for anything)
• Try these Kate Turabian, term paper, IEEE style
  for citations
• DONT FORGET THE LIBRARY!
• The MSU library has, for example, recent MS and
  PhD dissertations online

46
Some MSU Resources

• Departmental
• www.cs.msstate.edu
• http//www.cs.msstate.edu/GRADUATE/
• Graduate School
• http//www.msstate.edu/dept/grad
• Registrars Office
• http//www.msstate.edu/dept/registrar/

47
Acknowledgements

• All my students over the past several years
• Dr. Donna Reese for help with making many clear
  MSU/CS hypotheses out of murky ones
• Dr. Ed Allen for becoming the style guru for
  MSU/CS MS and PhD dissertations
• Mr. Puri Bangalore, Mr. Srigurunath Chakravarthi,
  and Dr. Rossen Dimitrov for specific input toward
  this talk
• All the professors who have taught the Graduate
  Seminars at MSU/CS over last decade