Rumination 31 March 2022

The last few days and the coming few days have been and are going to be very eventful. For starters , I tried to build a data aggregation platform for receiving driving condition data and driver emotion data with the help of some IoT sensors and our very own CNNs and also with the help of AWS. I worked on it for nearly a week and a half and didn’t sleep for an entire night. I had been begging my advisor to take a look for a number of days which he couldn’t respond to because he was busy. Well , the meeting the day it did happen was decent I guess , he however could not pay attention to the details , because he had to leave for calls repeatedly. :( :( . I thus could only give the demo to his subordinates. Aah well. He did indicate to me the research potential in my project , so I guess that was decent. Anyway.

I really need some good publications on my side, that are well directed. I have been aiming to publish in a really weird manner. From Server less Computation to Social Networking for the elderly to Healthcare Security to Driving environment Data Aggregation. I seem to be falling into my old pattern of not making any progress because of excess diversification. I don’t know , I guess I will risk it. Maybe I am possessed by some kind of spirit, pushing me in this direction.

I also haven’t yet gained completely, the ability to create ideas that are well researched and thought through. My ideas are often based on the experiences I have had in life, and not really on the basis of the existing scientific literature. Not that it is unacceptable in research , but for some reason I have a feeling , my work will be noticed a lot more if I was to base it off of the existing literature. It probably means I have to become really good at doing literature surveys. I have a few ideas in this regard that could be useful to someone trying to conduct a literature survey.

1. Frame a research question and come up with at least 50 keywords/keyphrases related to it. Spend at least a day coming up with them , because they are the most important part of your research
2. Go to the best publisher/platform in your field and query the keywords that you came up with and download 15 of the most highly cited journal papers. Then download 15 of the most highly cited conference papers in your domain. (Journal papers generally are more detailed and contain a deeper discussion of the research topic , whereas conference papers are relatively shorter and contain the results of smaller experiments)
3. Next try to come up with a classification scheme of the papers that you just downloaded. Now revise your keywords , with this new classification scheme in mind and repeat the process of downloading papers. For example if you are researching Quantum Computing for Blockchain, you could probably classify the papers you are reading into data transmission papers, security papers, computational complexity analysis papers etc
4. Download about 100–150 papers and simply skim through them, trying to figure out if they are worth reading. You can do this by going over its abstract , its results its methodology etc and shortlist about 45 papers, while maintaining your classification scheme.
5. Once you have shortlisted about 45 really good papers, place them in a citation management system such as the Mendeley Reference Manager, you and your research partner now have to go through these papers in an extremely detailed manner. To do this you should highlight portions of the research paper that really stand out to you, and also skim through the cited article included in the citation if you want deeper details about it.
6. Now these highlights can be of two types. There are some highlights that are common to all papers, such as the main research question in the abstract , the inferences in the conclusion , the implications etc, but there are some portions in the research paper that are not common to all research paper and will depend on your understanding of it eg those highlights that demonstrate the ingenuity of the authors, these must definitely be highlighted.
7. Once you have highlighted sections of the paper you have thought were important. You should construct a mind map demonstrating the connections between the ideas within the paper and the connections between the ideas across multiple papers. The mind map can also be used to serialize how you will organize the final draft of the literature survey. You can do this by numbering the nodes in the mind map.
8. For every paper you read , you must ask at least 10 questions that are your own, critiquing the methodology , the limitations in the results , the reproducibility , the lack of information. This information can be graphically represented in the mind map as probably a crack in a bubble or something.
9. You must also try to connect ideas in your mind map that aren’t directly connected. I have a feeling it is in this step that experiments are conducted. You will probably have to conduct your own experiments to find connections between the ideas in the mind map and re-conduct the entire literature survey for new information and then design new experiments and so on.
10. Once you carry this process out across multiple research articles you will probably have a very deep understanding of your research idea and will hopefully be able to create new knowledge in it.

The method that I discussed above isn’t absolutely new in any sense and is a mild variation of the traditional technique that is taught in Research Methodology 101. I have only introduced the mind map and visualization part. In most work that I have seen they don’t exactly address how to create a coherence between multiple papers in a literature survey and they simply hope that it happens out of sheer intelligence. However I have seen that this is often not the case because research papers can have very complex ideas in them and we are often only used to simply critiquing one idea in the presence of another and not critiquing a complex system of ideas with another.

A mind map detailing the interactions between ideas will probably go a long way in creating the necessary clarity for a researcher who wants to advance his field.

Ohh well , I meant to talk about something completely different today. Honestly I only wrote this post to run from the anxiety of having a deadline on a really difficult project.

I dont know why but researching about research is turning out to be kind of fulfilling.