HOW TO GET FROM COMPLETED EXPERIMENTS TO A COMPLETED THESIS ----------------------------------------------------------- TO KEEP TRACK OF DRAFTS OF DOCUMENTS, PLEASE ADD THE DATE TO THE NAME OF THE FILE THAT YOU SEND ME, e.g. thesis17Feb2004.pdf. To produce the final version of the thesis as efficiently as possible, here is the recommended process: STEP 1: GENERATE A PRESENTATION ------------------------------- Produce a presentation that completely describes the research. The presentation should follow the outline in Tim Brecht: "A simple approach to thesis writing" http://www.cs.dal.ca/~grad/ThesisHints.html and follow the guidelines on how to make a good presentation in: http://www.cs.dal.ca/~eem/6505/MLAdmin.html There is no time limit on your presentation. Use as many slides as necessary to describe the research. Typically 40-60 slides would be more than enough. I will work closely with you to produce a high quality presentation. The reasons for working on the presentation first are: - it is much faster for me to give you feedback on the presentation than the thesis itself. - it is much faster for you to put the presentation together, because it contains much less wording than the thesis. You may want to do some thesis writing as you do the research or work on the presentation. I recommend that such writing be short descriptions of specific segments of the research (for example description of your algorithm, experimental results, abstract). Avoid wasting time on the literature review section at this stage. FIGURES The recommended program for drawing figures is xfig (on Unix/X windows). It is freely available on the Internet, and it has been installed on locutus. If you are a Windows diehard, please use Corel Draw. If you want to use another program, please talk to me first. STEP 2: GENERATE THE FIRST THESIS DRAFT --------------------------------------- Once a high quality presentation is finished, glue together the various thesis fragments you have using the presentation as the guide. The ideas in the thesis should flow in the same way as in the presentation. It is important to motivate each part of your thesis before you present the technical details. Material that is too detailed should go into appendices. The main body of the thesis should be pleasantly readable without too much head scratching for the reader. STYLE HINTS: - It is poor style to start a sentence with: Figure 10 contains xxx ... Equation 10 shows xxx ... Table 10 shows xxx ... Rephrase as: xxx, as shown in Fig. 10, Eq. 10, Table 10... - When you make references to figures or equations, use Fig. 10, Eq. 10 instead of Figure 10, Equation 10. - When presenting experimental results, use a logical order, not a chronological order. The thesis should not read like a diary (I first did x, then I did y, ...). - Figure captions should provide a stand-alone description of the figure, to allow someone skimming the thesis to get the gist of the thesis without reading the detailed text. It is acceptable to duplicate some of the text in the figure caption. - When considering how much detail to include in the thesis, keep in mind that the reader should be able to reproduce your results based on the information in the thesis alone. - For the previous work/background section, stick to the few articles that are most closely related to your work. There is no need to survey the entire field. - Run a spell checker! I will return theses that have spelling errors that can be caught by a spell checker. - If you would like to send me comments about your thesis, please use the \comment{} command in latex, or annotate the pdf file you send me. It is less convenient to send me a separate email, and even less convenient to send me a MSW file. - Generate and send me a postscript or a pdf file from the LaTeX file. LATEX HINTS - Equations and tables should be properly formatted in the LaTeX source file to maximize readability. - In eqnarray's and tables, column delimiters "&" should line up across lines. - Do not use more brackets than required! * If the superscript or subscript consists of a single character, you do not need brackets around it: a_{i} is the same as a_i \sum_{i} is the same as \sum_i * Items in itemize and enumerate environments do not need brackets around them. * cell contents in tables do not need brackets around them - To prevent the reference from appearing in the beginning of a line, do not separate the \ref{xx} or \cite{xx} from the preceding word by a space or new line in the .tex file. Use instead a tilde: blah~\cite{xx} instead of blah \cite{xx} - To prevent indentation of the line right after an equation, do not leave an empty line between the equation and the previous text in the .tex file. STEP 3: COMMENTS ON A THESIS DRAFT ---------------------------------- Comments are typically of two kinds: - Minor editorial comments: grammatical, syntactical and minor rewording corrections. These comments typically take minutes to address. - Major comments: They involve rewriting of paragraphs, clarification of claims made in the thesis, request for additional text. These comments may take hours or days to address. I start the text of major comments with ==> or *** and may number them (1,2,3,...) if there is more than one such comment in a single page. You can go to the location in the pdf file to which the comments refer by clicking on the comment in the list. -------------- Acrobat annotation types ------------------------------ I use the following annotation tools in Acrobat: Tool Appearance My use of the tool ----- ------------ ------------------ Note: note icon. (*) Major, general comments, important. Pencil: blue handdrawn line I circle corrections previously explained or obvious spelling errors. Line: blue straight line (same as underline) Highlight: highlight of text (*) corrections, rewording, comments associated with highlighted text. Major comments start with ==> Underline: red underlining of text important point, for my information only. Strikeout: red line crossing out text delete that text. --------------------------------------------------------------------------- (*) All annotations can be associated with a pop-up text window, that appears if you double click on the annotation. You can cut and paste from pop-up window. I only associate text with note and highlight annotations. You can also view the list of (and visit) comments by selecting in Acrobat: Window -> Comments Double click on entries of the list to visit the comments and the associated text. The comment list can be sorted in different ways. You can also view the start of the text of the comments (which is ==> for the major comments associated with highlighted text). STEP 4: REVISING THE THESIS AND RESPONDING TO COMMENTS ------------------------------------------------------ Include the following macros in the preable of your LaTeX file (before the \begin{document}) \newcommand{\comment}[1]{ {\tiny {#1} } } % the regular %%\newcommand{\comment}[1]{ } %hide stuff, for final version \long\def\invis#1{} %make text invisible \usepackage{color} For each *major* comment *only*, do the following: a) In the text, please include a comment of yours: \comment{*** ...your response to my comment in the revised draft...} The comments will appear with tiny letters. - Text that you added/substantially modified to address the comment should be wrapped in: \textcolor{red}{...new text...} This will make it appear red. and enclose the *old* text in: \comment{oldtext} before you process the file. Text that you think might be useful in subsequent revisions can be made invisible by enclosing in \invis{...}. For the final version, you can eliminate the old text by activating the second definition of \comment above, and use search and replace to turn red text to black. If time allows, remove all \textcolor and \comment calls for the LaTeX file. b) Run the LaTeX on the text. Your explanations in tiny and the new text in red will greatly speed up my review of your next draft, as I will not look through the whole thesis, but to just at how you addressed the important comments. For the final version, delete all \comment{...} and \textcolor{red}{...} entries. c) We will iterate between Steps 3 and 4 as many times as necessary to produce a thesis of acceptable quality. Typically 2-3 iterations would be sufficient (the first major, the other two minor).