This course deals with computer science (CS) aspects of social network analysis (SNA), and is part of the master computer science programme at Leiden University.

« Back to SNACS course webpage

Course project paper

Papers on a particular topic all consider different algorithms or methods for solving roughly the same problem. The goal of the project is to make a comparison of at least two of these papers based on experiments using real social network data. In the paper the different techniques are analyzed and compared in detail using extensive experiments. The paper, to be written in LaTeX, has to follow the format of an actual scientific paper. If you do choose to use an other format than LaTeX (which is highly discouraged), then you are fully responsible for making the final result look as professional as a LaTeX paper.

A template snacspaper.tex is available, which you can compile with pdflatex. You also need this .cls-file and .bib-file (which you can compile using bibtex). We are using an adjusted version of the 2-column ACM conference proceedings template. Of course, feel free to use any additional LaTeX packages. Likely, you will need tikz, graphicx, etc.

Your paper should be structured like an actual scientific paper and should be divided into various logical sections and possible subsections using at least 6, but no more than 10 two-column pages. Your paper could look something like this:

Abstract (0.5 column)

  • In 4-10 sentences, summarize the paper.
  • What is the main problem studied, what techniques are proposed and why, how well do they work and what is the main conclusion?
  • Write this as the very last thing.

Introduction (1-2 pages)

  • What is the main problem that is going to be be studied in this paper? (in words)
  • Why is this problem important?
  • What applications does the problem have?
  • What are existing techniques for solving this problem? (in words) Perhaps already provide some references to previous work.
  • How is this paper going to contribute w.r.t. existing techniques and related/previous work? Spend a section on this later in the paper if needed.
  • What does the rest of the paper look like? In one paragraph, list the topics of the following sections.

Problem Statement (1-3 pages)

  • Give definitions, possibly soome theorems, explain nontrivial notation, etc.
  • Formal problem statement, sub-problems, etc.
  • What about space and time complexity?

Related and previous work (0.5 page)

  • What related work is out there? Provide a good amount of references.
  • How does this paper differ from related/previous work?

Suggested Approach(es) and Algorithms (2-4 pages)

  • In one or more sections explain proposed techniques that are being compared. Reference other works properly.
  • Discuss differences between previous suggested approaches and do some suggestions to combine techniques.
  • Provide schema's, pictures (using the tikz environment?), etc. to clarify proposed techniques.
  • Perhaps include some pseudo-code using the algorithm(ic) environment?

Datasets (1 page)

  • Which data sets are you going to use?
  • Why is this data good for these experiments?
  • Is the data possible biased and how does this affect the experiments?

Experiments and results (2-4 pages)

  • What experiments can be used to compare, test and verify the suggested approaches?
  • What do you measure in each experiment? Quality, running time, error size? Be precise.
  • What are the main results?
  • For datasets that perform either very well or not so well, try to find out why.
  • Zoom in with additional experiments on noticeable results.
  • Can you relate the performance of the algorithms to some of the properties of the datasets? Or: can you define for which particular datasets a certain technique works well, and why?
  • Report on experiments using tables and figures. Probably, you will need to use gnuplot. Remember captions and axis labels.
  • Make sure that the type of data/result that you want to communicate is suitable for the chosen presentation mode.

Conclusion (0.5 page)

  • What was our problem statement again? (1 sentence)
  • What was learned from the experiments?
  • To what extend can the problem be solved?
  • What works, what does not?
  • What remain open problems?
  • Can you give suggestions for future work in this area?

References (max. 0.5 page)

  • Provide a list of complete and precise references.
  • For an article, provide author, title of work, journal- or conference name, page numbers and year.
  • For a book, provide author, title, press and year
  • For a website, provide author, title, URL and data accessed.