Research Relay

• 

  Picking a topic via discord bot

• 

  Computer Graphics article

Inspiration

We were inspired by the vast and ever-growing collection of research articles in databases like arXiv. It's easy to get overwhelmed by the sheer volume of new papers, so we thought, why not build a solution that brings the most relevant research directly to users based on their preferences? We wanted to make accessing and engaging with cutting-edge research as easy and interactive as chatting with a friend.

What it does

Our project is a Discord bot that connects users with research articles from the arXiv database. Users can choose topics they’re interested in, and the bot sends them daily or on-demand summaries of relevant research. The bot can also engage users in conversations about the articles, asking questions or clarifying content. Additionally, users can like or dislike articles, allowing the bot to refine recommendations over time.

How we built it

We used the arXiv API to retrieve research articles and Python with Natural Language Processing (NLP) tools to generate concise summaries. The bot is built using Discord’s API and integrates with a database to store user preferences. We also employed machine learning algorithms to help the bot learn from user feedback and provide more tailored content over time.

Challenges we ran into

One of the major challenges was fine-tuning the summarization process. Research articles can be dense and highly technical, so creating concise yet informative summaries was tricky. Additionally, integrating the user feedback loop and building an effective recommendation system presented a challenge, especially with varying user preferences across topics.

Accomplishments that we're proud of

We’re proud of building a functional bot that not only retrieves and summarizes research articles but also interacts with users in a meaningful way. The ability for the bot to refine its recommendations based on user preferences is something we’re particularly excited about. Also, incorporating a conversational element to discuss articles is a big milestone for us.

What we learned

We learned a lot about how to work with APIs, especially when pulling data from research databases like arXiv. Additionally, building out the recommendation engine taught us more about machine learning and user profiling. We also gained experience in NLP and how to simplify complex academic writing into digestible summaries.

What's next for Untitled

Next, we plan to enhance the bot's ability to personalize recommendations based on user feedback by improving the like/dislike system. We also want to make the conversation around articles more dynamic, so the bot can dive deeper into discussions or offer insights based on user queries. Finally, we aim to expand the bot’s capabilities to other research databases and create more interactive experiences for users.

Built With

• cohereapi
• discordapi
• llamaindex
• python