Tips and Tricks

Online Latex

In research, it is popular to write papers using latex. Something like Microsoft word may give you hard time. So, go ahead with latex. Yes, some leaning curve, but after that it will be much easier.

Though you can download and use a latex program, but is much better to use online one, especially the ability to share it with others. I used to have a shared project with my supervisor to get things done. You may try Overleaf service.

Paper BibTeX

In latex we use the bib file to list our citations. It might be annoying to keep writing the bibtex by yourself. Here is a trick. Search dblp site with paper name, find it and from its menus click bibtex. Just copy the file :). Most of the time, the result is accurate, though sometimes they list the paper but without the correct conference, so you may need to double check. One way for verifying is googling behind the paper, or accessing author website or check Google Scholar, get some paper cited the work and see what they wrote.

Finally, it might be good to use a uniform way for the bibtex entry. My Favourite way is LastName_Conference_year_keyword. For example, IbrahimECCV18RelationalNetwork , IbrahimCVPR16Deepactivity.

You may also clean it from the many entries it have to a simple one (I guess all these exact details were important in the past). Here is my paper example:

@inproceedings{IbrahimECCV18RelationalNetwork,
  author    = {Mostafa S. Ibrahim and Greg Mori},
  title     = {Hierarchical relational network for group activity recognition and retrieval},
  booktitle = {2018 European Conference on Computer Vision (ECCV)},
  year      = {2018}
}

Writing Related Work Section

There are a few things I want to mention here.

Notes on writing this section

  • Start with determining what are the different areas or problems that are related to your problem
  • Build a list of the related papers for each aspect (see next point)
  • You don't just list the papers. You need for some of them to contrast your work against them stating why you are different

Determining the list of related papers

  • You may need to explicity check the conferences in the last few years
  • Assume you know initial X papers that are highly releated
  • Find who cited this paper from Google Scholar
  • For each title that seems close to the paper goal, check it if it is also releated
  • Citing the most related popular papers is important. A reviewer maybe upset that you are not aware with important work
  • Citing the recent related work is nice thing and give them some visibility.

Summarising the papers

  • Coming to the boring part :)
  • Probably you know some of the work, but not of all of them. You maybe running out of time and wish to get things done so fast!
  • Here is a shortcut for this case.
  • Say you want to write a summary statement about paper X
  • Give a read for the paper abstract and its contribution section. Maybe also check their figures.
  • Use Google Scholar to get the papers cited X. Find up to 5 papers that really wrote a summary of this paper.
  • Write down these statements in a file. Now you know how papers perceived this work or what are the most important aspect from their perspective.
  • Use the paper abstract and these summarise to paraphrase a new statement about the paper.