The problem Insightly solves

No matter what career field you’re in or how high up you are, there’s always more to learn. Research unlocks the unknowns, lets you explore the world from different perspectives, and fuels a deeper understanding. Reports state that 90% of ideas fail because of lack of research and proper background study. Nowadays for most researchers in academia, reading through different papers has become a hassle. However, it's hard for us to get insight on the same because they are very long and time-consuming to find. This inspired us to create Insightly, a smart webapp where one can effectively analyse research papers and understand key points packed with multiple other AI-based features which you'll follow through the post.
But wait? why “Research”?

Research is pretty much the base of every hackathon project that one has ever done & will do. But apart from this, let’s discuss the top 10 points for why one should do research.

1. Research expands your knowledge base.
2. Research gives you the latest information.
3. Research helps you know what you’re up against.
4. Research builds your credibility.
5. Research helps you narrow your scope.
6. Research teaches you better discernment.
7. Research introduces you to new ideas.
8. Research helps with problem-solving.
9. Research helps you reach people.
10. Research encourages curiosity.

Insightly is a productive tool that is used to efficiently examine research papers, without needing to read through the entire paper. It streamlines easy access to published papers and significantly reduces researching time. With this platform, the user can:
View summaries of the research paper(s)
Find tagged topic headers for the paper
Get more info on research topics through Q&A
Get sentiment status of the topic
See similar paper recommendations
Personalized Profiling

Challenges I ran into

I had to brainstorm and come up with a feasible idea that was innovative and could be implemented within the given time frame.
I went with a minimalist approach by building the front-end with HTML, CSS & Vanilla Javascript. After OAuth-ing with Firebase, we land on the homepage, where we can leverage a bunch of useful tools according to our needs.

We can upload our own pdf document or can search through papers available on ArXiv or via custom URL. Once we do that we have the option to generate total or partial summary of the selected which is powered by our own Custom Trained Superfast Machine Learning model. Moreover, we can also analyze the sentiment of the same if we require. Furthermore, using BigBERT we are also able to resolve any sort of questions taken as an input promt to return accurate answers. The backing is running on splitted pipelines of Pytorch with OnnX. And last but not least, all these datas shall be stored in the user dashboard utilizing which, the machine can automatically recommend more similar papers via cosine similarity.

Research 📚
Research is the key to empathizing with users: we found our specific user group early and that paves the way for our whole project. Here are a few of the resources that were helpful to me —
https://honisoit.com/2022/04/the-curse-of-knowledge-why-are-academic-papers-so-difficult-to-read
https://scienceandword.com/why-are-research-papers-hard-to-read
https://www.ucd.ie/all/t4media/Chapter%209%20It%20is%20really%20difficult%20to%20read%20scientific%20papers%20-%20John%20O%20Connor.pdf
https://thewire.in/science/scientific-study-says-science-papers-become-harder-read-last-century
https://sheridancollege.libguides.com/researchmethods
https://blogs.warwick.ac.uk/inspireslearning/tag/methodology
https://www.london.ac.uk/courses/understanding-research-methods
https://researcher.life/blog/article/reading-research-papers
https://goldbio.com/articles/article/how-to-read-and-understand-hard-scientific-papers