How to perform Computing Research
- Our Library has the two main databases you will need:
Cite SeerX is also a tool that has been around longer than Google Scholar and is an excellent tool with more in-depth information like graphs that show the citations by year.
DBLP is a good source to see a specific author or conference etc.
- BibTeX tools include:
Having the tools is only a small part of the process of researching. In the following sections I'll address how to find papers, how to read them once you have found one, evaluating the paper and deciding if it should be in your bibliography.
Many people ask me about using Wikipedia or other online sources in their Bibliography. For my classes, I give the following answer: Although I do not consider wikipedia a bad source of information, (I do consider it a good place to start looking for articles) research papers base their value on a peer-review process. Hence research papers should not use non-peer reviewed material as authoritative sources. This means, you shall not use Wikipedia as a source in your Bibliography! Instead, use Wikipedia as a resource in your research.
How do I find papers?
A classroom demo of using these tools is available here. But that is only step one. Paper titles and abstracts will determine weather you will save the paper (and the BibTeX that goes with it) or discard it. You should finish searching when you have found 15-20 promising papers. The next step is to read them...
How do I read a paper?
Re-read the abstract, then read the introduction and conclusion. Highlight and annotate as you read! The introduction is the author's chance to convince you that you should read the rest of his/her paper! Papers that don't make this cut will not be fully evaluated. If you find yourself with fewer than 10 papers, you may need to go back and search for more papers. Papers citing the ones you keep are prime targets for addition. Also seminal papers that are cited in the papers you keep should be given serious attention (seminal papers have lots of citations).
How do I evaluate a paper?
In order to evaluate a paper, you must first determine what contribution that paper makes. Reading the introduction and conclusion should come close, but you need to read the entire paper to understand how the authors created the contribution. Is it trivial? You might think so, but it may be a brilliant insight. If the brilliant insight sounds simple, they have done an excellent job of explaining it! How do they analyze their contribution? Is the analysis similar to other papers you read? If not, why not? Is the contribution based on previous work? Again look at previous work. I once reviewed a journal article that included verbatim contents from three conference papers, but included only a few paragraphs of "glue" text to hold it all together. Plagiarized text does not a good paper make! However, journal articles are usually summaries of past work and will contain some previous work verbatim.
Your notes should restate, in your own words, the author's contribution. These usually show up in your related work or introduction section. Writing your own summary often helps later on when you are assembling your paper.
Is this paper relevant to my topic?
By now this should be obvious. The questions you should answer include: Is this work preliminary to what I am doing (not common for undergrad students)? Is this work preliminary to industry standard software or practices (common for all), Is this something that I am writing an implementation for in a class (very common for undergrad students)? Answering these question will help you decide where in the paper to place the references. Intro/related work/body of the paper respectively.
The guidelines do not comprehensively cover research, but give you a place to start. There always exist exceptions to the guidelines and you will no doubt experience aha moments about how the process actually takes place. If you experience one of these moments and think it would help others along the way, shoot me and email and I'll consider putting it up on this site.