forest

Inspiration

Our team was inspired by how overwhelming it can be to navigate academic research. Reading multiple papers to identify connections is time-consuming, so we wanted to build a tool that automatically summarizes papers and visualizes their relationships, making research exploration faster and more intuitive.

What it does

Forest is a Chrome extension that extracts paper information from arXiv and Semantic Scholar, summarizes the content using GPT models (or open-source alternatives), and organizes papers through semantic clustering. It then displays the connections in an interactive knowledge graph, helping researchers see how papers relate and discover new insights efficiently.

How we built it

We built Forest as a Chrome extension using Manifest v3 with a popup UI. Content scripts extract data from paper pages, PDFs are parsed with pdf.js, and NLP models generate summaries. Papers are clustered using SBERT embeddings and visualized with D3.js to create an interactive knowledge graph. Our team collaborated closely to integrate each component and ensure smooth communication between the popup, background scripts, and content scripts.

Challenges we ran into

Coordinating between multiple components of the extension—content scripts, popup, background scripts, and NLP processing—was challenging. Parsing PDFs programmatically and ensuring efficient clustering required careful optimization. Additionally, running large NLP models for summarization and embeddings posed resource constraints, so we had to design the workflow to be lightweight while still functional.

Accomplishments that we're proud of

We successfully built a fully functional Chrome extension that summarizes research papers and visually maps their relationships. Our semantic clustering and knowledge graph visualization provide a unique way to explore academic literature. The project demonstrates a practical integration of NLP, browser extension development, and interactive visualization within a short timeframe.

What we learned

We learned how to work effectively as a team, combining different technical skills to create a cohesive project. Technically, we gained experience in Chrome extension development, NLP for summarization and embeddings, semantic clustering, and interactive graph visualization. We also improved our workflow coordination and debugging across multiple interconnected components.

What's next for forest

Next, we plan to expand Forest’s capabilities by enabling full-text PDF analysis, integrating more open-source LLMs for local summarization, and improving the knowledge graph visualization with additional interactivity. We also hope to explore features like automated citation suggestions and personalized research recommendations to make Forest an even more powerful research assistant.

Built With

• amazon-web-services
• d3.js
• gpt
• html/css/js
• manifest-v3
• oss