My name is Vedha Krishna Yarasuri. In this blog, I will go through the entire workflow of project development to writing a research paper.
This blog consists of the following blocks,
- What is a research paper?
- What is the use of a paper?
- Steps involved in the whole process.
- Last words
What is a research paper?
One may think that a research paper involves in-depth focussed research work. But it’s not the case, A research paper is yet another article, but the difference is it looks much more professional, precise, and accurate. The author has to use a minimum number of words to convey the entire idea (Being precise and accurate).
A research paper is an idea proposal shaped with results. The details presented in any paper are validated and tested. The idea may be a small technical hack but it should have results. The quality of the research paper is based on the idea of the researcher. If the idea is a new technique then it would grab a lot of attention.
What is the use of a paper?
As discussed earlier, a research paper consists of a validated idea. Once the paper is published, then anyone in the world can access it. So It means someone’s work may help another person to shape their own idea. Since it is not just a hypothesis, it may also help a reader to get to a possible solution or it may end up in new techniques based on someone’s work.
For example, When I have participated in a few hackathons, I used to validated my idea with the help of someone’s previous work in the same domain and came to a conclusion. The main motive behind writing a research paper is to present a possible idea to the world so that someone in the same domain may get help and create something new or extended work.
Steps involved in the process
Writing a paper is just an extension of our work. Working on a research paper is of 2 phases.
Phase 1: Working on the project
Phase 2: Paper Writing
1. Choosing a Mentor and technology.
2. Getting Idea.
3. Past works.
4. Idea Validation.
5. Working on the Project.
7. Validating the results
Choosing a Mentor and technology
The first main step is to approach a proper mentor in our domain. If someone is new to any technology then I would suggest them to pick a technology that is fascinating to them and find a mentor in the chosen domain.
The mentor would have a lot of experience in both writing a paper and working on a project. He would help us if we stuck in the workflow. Also when we are new to writing a paper, we would make a lot of errors. When I was writing my first paper, My mentor had removed more than 90 percent of my writings in my first attempt. At that time I was not precise and had a lot of plagiarism errors. Later upon 3–4 attempts, I made fewer mistakes and understood what it really means to be precise and accurate.
This part of the project development is more difficult than others. This is because getting an idea is a difficult process. In my past, My team has only worked on getting an idea for a whole month. It’s completely based on our brain. In my experience, I have found out that getting an idea just requires remembrance of our problems or our surrounding problems. Thinking in that way is the easiest way to get an idea.
For example, When I was writing programming functions related to pre-processing for machine learning, I found out that it was somewhat difficult as there would be so many functions for handling and manipulating the data. This was my problem and I thought this would be again a difficult procedure for beginners in the field of machine learning, so I have compiled all my functions and made a package named flow2ml and released it as an open-source. Now for me working on that part would be easy. So I have created a project idea named flow2ml from my problem.
In this phase of project development, we can find various resources to check various other implements done by several researchers. We can find them either in any of the medium blogs or IEEE or Scopus or youtube or in any other internet material.
For example, In one of my projects aimed to solve the problems of farmers, I have searched several articles and materials which are helpful in building our project. In this process, I have found various other implementations as well. I have found ways on applying various machine learning algorithms to predict various sensor values in the next time stamps and I have also found information about Hyper-Spectral image analysis, utilizing Bhuvan data which I never thought of implementing.
Now, someone’s past work helped me and my team in expanding our present idea to a great extent.
This is another important step in project development. This may extend our idea to a great level, where our idea may help in solving some real-world problems and get patented. After a small research on the past works in our domain, we can now have corrected ideas or added ideas whole as a big project idea.
In this stage, the developers try to search how does the main idea help to solve even a small problem or how does the idea impact society.
In my experience, I observed this step to have a lot of importance in hackathons. In view of paper writing, this part is important to shape our idea. By referring to various papers we can confirm that our idea is practically possible.
Working on the Project
After having a clear picture of the project idea, the next task is to start implementing it and make the hypothesis into practicality. In this stage, the mentor would help in providing a few useful resources using which we may get started.
We may face so many difficulties while doing the work but try to see everything from the perspective of the prototype on which everything would work. This makes debugging easier.
This part is the most interesting and important part of the whole process. If the results and hypothesis converge, then it seems everything looks good. If not then there might be any errors in implementing the previous parts. These results play an important role in both project work and paperwork.
Validating the results
After obtaining the results we need to validate them and prove that everything is going perfectly. We may again refer to a few papers to check if the results are okay.
Here the project mentor plays a crucial role, he/she may validate them with their experience and encourage you to make them the best by tuning a few parameters on which the entire project relies.
These results are the actual results that need to be put in the research paper.
This stage may lead to other stages where we would test the application in the real world and check its lifetime and market value and go further to make it into a product.
Now that we have completed working on our project. We can now extend to write a paper if the project involves proper techniques and results. At this point, the mentor helps in getting started with writing the paper.
3. Related Works.
4. System Architecture.
6. Results and analysis.
There are several stages in writing a paper. The first part of a paper is the abstract. The abstract contains a brief idea of the project. It gives an idea of various methods, software used, and various types of results taken while developing the project. This part gives an idea of what the paper contains.
With the help of this many readers can pick up the proper papers useful to them. There is a small section below the abstract for specifying keywords in the project which is about 2 lines. The abstract is written in bold letters.
This section introduces the readers to the project. It discusses the problem statement, its importance, its impact on society. On further reading, it will introduce the readers to various techniques and implementations involved in their work.
During the idea development phase whatever papers we have referred to would be listed out in this section. The author should give a small summary of what the referred paper conveys. By this we are able to cite various papers, this in fact helps a lot.
It helps the actual authors whose work has been referred to. The more the number of citations the more would be the value to the paper. Through this section, the authors of the paper can recommend their readers about what all the fascinating work is happening in the world. In the future, someone may cite the present paper through which the present work would get value.
This section gives the top-level view of the entire workflow of the work. This section contains the block diagram of the project. The complete pipeline will be discussed in this part. If the project involves any mathematical equations then they are presented here. For example in a machine learning project, The details such as data filtering, data augmentation techniques, type of ml techniques used will be discussed.
In this part, since we discuss the overall procedure, it helps the readers to relate our work with their work and let them empower their idea.
In one of my previous projects in deep learning, this part helped me in navigating to various open-source datasets. From this section, I have gained a lot of understanding of the higher-level view of the deep learning pipeline for large scale projects such as OCR, image captioning, etc.
In the previous section, we will give a top-level view of our project. In this section, we would go further in-depth about the main part of the project work. Many great innovations described the theoretical working of their works in this section. In machine learning, various theoretical concepts are first described here and later were proved in the results section.
For any hardware project, we would discuss the main hardware architecture, for machine learning projects we would discuss the algorithms used and how they contribute to achieving greater results.
This section lets the reader go a step further in understanding why the proposed architecture is best suitable for getting greater results. For example, in image classification, this section helps a lot in choosing the best architecture for the task.
Results and analysis
This section contains the results and analysis of the project work. It is always good to have various parameters for obtaining the results of the project. If we have more parameters to judge our prototype, then we would be able to see our prototype from various angles. This helps us to check the performance of our prototype. Each parameter would describe the prototype in various ways. If there is any small error in modeling then that may be pointed out here. This helps in improving the prototype further.
In the paper, this section holds a greater weightage. We can convey our results in the form of tables, graphs, images. It would be good if we can provide as many details as possible regarding our work. This helps the reader to relate methods and their outcomes. If the results obtained are best, then the reader would try to replicate the same. This section is proof of our description in the previous sections.
In the past in the field of deep learning various models such as inception, resnets, and other great models became so popular based on this section. This section proved many inventions to be the best.
This section contains the last words regarding the author's work. If the author has compared several methods then he can give a conclusion of the best performing method. The author can describe the future scope of their work. This section contains the entire story of the project in small words and will end in a small paragraph.
This is the last section of the entire paper. The author needs to mention all these references in this section. In the related works section, the author will mention the work done by various pioneers in small sentences. In this section, he needs to cite their paper. Through this, the reader can able to approach their papers. In general, there should be 15 references.
This brings to the end of the discussion.
Thank you readers,
I hope that this article will help you in building your own projects and papers.
I am an undergraduate student. So whatever I have mentioned in this blog is what I have observed from my little experiences. If there are any mistakes, then please mention them in the comments.
Vedha Krishna Yarasuri