Search techniques for finding references at the start of reviewing the literature for your research?

Question

Whenever we start new research, we need to do a literature review first in order to both get acquainted with the context and the concepts related to our research as well as to situate our research within the existing knowledge.

The problem is that in order to find relevant literature, you need to express what you are looking for in meaningful search terms. The more efficient and robust you make your "search string" (aka the keywords you use) the better you will express your information need and find relevant references. If someone is looking for "text classification for categorising books", it is better to word it as "book genre identification".

Are there any techniques for finding the perfect wording for your initial literature search? Myself, for example, I try to find an initial paper that is relevant and the browse through its cited papers to see if there is any paper that describes what I am looking for in a better way.

Answer 1

You don't need "perfect wording for your initial literature search". Getting started on a new project is an iterative process. You think about the problem. You read some papers. You think some more. You find new places to look. You have ideas of your own and follow them up. Eventually you will discover things that are not yet known.

When you are pleased with your progress you do the literature review that's appropriate for the paper you are writing.

Answer 2

There is no one approach. Presumably, you know enough about the topic to find one seminal paper that's been heavily cited. You then look at the heavily cited papers the cited your first paper. You then go read the other important papers that those authors cite.

Long story short, pick an interesting paper, and search forwards and backward in time from that point.

Answer 3

As @EthanBolker said, a good literature search is an iterative process that evolves along with your research progress. The very notion of a "perfect search string" from the very beginning constrains you to think only in terms of what you knew when you started the project without adjusting for your continuous learning as you advance in the project.

I would further add that it is a widely mistaken concept that the primary purpose of the literature review of a scholarly article (which I call a "background literature review") is to summarize the related literature. Well, that is often accurate in practice, but in my opinion, that is not the most useful thing that it might do. Rather, the primary purpose of a background literature review should be to clearly carve out the novel scholarly contribution of your article in relation to the literature. That is, by the time you have completed the article, the background literature review should focus on clearly showing how the existing literature is related or similar to what you are doing, but how your article goes beyond the existing literature and makes novel scholarly contributions beyond what exists.

You cannot write such an effective background literature review at the beginning of your project when you yourself are not quite clear what is the extent of your contribution. Although it is certainly good to do an initial search before you start so that you have an idea of the research landscape, it is only by the time you are finished and you can properly appraise what you have done that you would be able to complete a background literature review that properly places your contribution in context of all relevant literature. So, from that perspective, there cannot be a meaningful notion of a "perfect search string" at the beginning of a research project.

I think what you have described is a good way to start: "I try to find an initial paper that is relevant and the browse through its cited papers to see if there is any paper that describes what I am looking for in a better way." In addition, you should probably use Google Scholar to do a forward citation search, that is, find articles that have cited your initial papers and see if something more recent is related. Then, after you have completed your project and have a good idea of what your contributions are, you can try to search on the keywords based on your contributions to find if anyone has already done anything similar. Then read such articles to clearly understand how your contributions are different so that you can then write a background literature review that clearly highlights the uniqueness of your contributions.

Answer 4

Presumably you will want to start a new type of research after:

1. You met someone that does some research that interests you
2. You read a paper that you liked and it sparked a new interest
3. You saw a talk about a subject that you liked

If 1. just ask that person, if 2. read carefully the introduction of said paper and list all the references cited there. Then go to those papers and read those introductions and do this iteratively. After a couple of iterations you will see some papers appear all the time, those are the seminal papers. Read those, then you have an idea what to search for.

If you are lucky, somebody has already written a review, so it's almost guaranteed that after reading a couple of recent introductions it will pop up there.

If 3. you can look for papers that they wrote and start from there. Perhaps you can even ask them to give a talk in your university (or ask a professor to invite them).