Scientific Paper Discovery

Arumoy Shome

2021-06-09 Wed 16:26

In this article I describe my process for discovering relevant and important papers in a new scientific field. Since I am working towards a doctoral degree, I need to read a lot to develop a deep understanding of my field of research and stay current on the latest developments.

Where to Start

The starting point is always the most difficult. If possible, the best place to start is through recommendation from your peers/supervisor/advisor. If this is not possible, then coming up with a few good keywords from your research question is the next best approach. Reading a few non-scientific work (such as wikipedia) might be necessary to obtain the “buzz words” of technical jargon which are required to formulate the search queries.

The right way to familiarize oneself with a new field is through a systematic literature review. However, I find that to be too big a commitment for a new field of research (I may not stick with it if I don’t find it that interesting). So the smarter (and quicker way) to familiarize oneself with the literature is to read the related papers.

Finding More Papers

I categorise this search process into two parts: 1. Intrinsic and 2. Extrinsic.

We can find relevant papers from within the current paper we are reading. This can be done by first reading the related work section of the paper and second by reviewing the reference section of the paper. The related work section often leads to papers that are at a similar level of technical depth as the current paper. The papers in the reference section tend to be more general and provide a good high level overview of the field.

We can also find relevant papers from the search engine (Google Scholar in my case). The cited by feature of GS is a good place to find forward references or the papers which are citing the current paper. These tend to be more specific and bleeding edge as they came after the current paper. The related articles are also a good place to find similar bodies of work. Google Scholar and Connected Papers are good resources to explore this.

Usually, 5-10 good, well reputed papers is sufficient to start with. And then, reading their related papers (and in turn their related papers, so on and so forth as necessary), should result in a collection of bibliography which paints a good picture of the field.

Staying Current

One also needs to stay current on the field. For this, Google Scholar alerts are a good start. By now the “position papers” and important authors in the field should be known. Following these authors on Google Scholar and creating an alert for when a position paper gets cited is a good way to stay up to speed. Another approach is to keep an eye on the relevant conferences and find papers directly from there.

Concluding Remarks and Acknowledgements

I was surprised to find that not a lot of information on this matter is easily available. It is similar to how students are not taught how to learn effectively, before they start attending university classes. The emphasis on the “meta”, the methodology of doing something is lacking. I cannot take credit for the wisdom above, a lot of it came from conversations with my peers and Google Scholar help website.