ResearchLens

• 

  Effortlessly discover, summarize, and understand research papers with AI-powered insights—designed for students, by a student.

Inspiration

As a research student, I know the struggle of sifting through long academic papers, even when the topic is exciting. Sometimes, you simply don't have the time to digest all that information. VectorScholar.AI was created to solve this problem. It helps you quickly find relevant papers, generate concise summaries, and understand complex research topics more effectively. Whether you're looking for similar studies, want insights at a glance, or need to stay on top of a specific field, VectorScholar.AI supercharges your research workflow by providing AI-powered recommendations and even converting text to speech for on-the-go learning.

What it does

ResearchLens is an AI-powered platform that helps researchers quickly discover, summarize, and understand research papers. It allows users to upload papers and provides a suite of features such as finding similar papers, generating concise summaries, offering recommendations for related work, and converting text to speech. The goal is to make research more accessible, efficient, and manageable, especially for students and academics dealing with long and complex research articles.

How we built it

We built ResearchLens using a combination of React for the frontend and Flask for the backend, powered by InterSystems IRIS Vector Search. The backend leverages vector search capabilities to find similar papers and generate relevant suggestions. We also integrated OpenAI's language models for generating summaries and performing question-answering tasks. On the frontend, we designed a user-friendly interface for paper uploads, summaries, and text-to-speech features, enabling a seamless user experience.

Challenges we ran into

Frontend-backend integration: We faced issues connecting the frontend React app with the backend Flask API. The biggest hurdle was configuring proper API calls and handling responses. Handling complex data formats: Research papers come in various formats (PDF, TXT), which required us to implement proper parsing and text extraction methods for accurate processing. Vector search setup: Setting up and configuring the InterSystems IRIS vector search database was challenging, especially when fine-tuning its capabilities to return relevant results based on the similarity of research papers. Time constraints: Given the time limitations, ensuring the integration and deployment of all features without bugs was a constant challenge.

Accomplishments that we're proud of

Seamless AI-driven summaries: We successfully integrated OpenAI's models to generate concise and meaningful summaries of research papers, which can help users understand papers faster. Vector search capabilities: We implemented a vector-based search engine using InterSystems IRIS, which can provide relevant recommendations for similar papers, enhancing the overall research discovery process. Text-to-speech functionality: The ability to convert text summaries into speech adds a new layer of accessibility, making research easier to consume on the go. User-friendly design: Despite the technical challenges, we created a clean, intuitive interface that allows users to quickly upload papers and get results.

What we learned

APIs and integration: The project helped us strengthen our skills in integrating frontend with backend APIs, especially handling file uploads and managing responses. Vector search technology: Working with InterSystems IRIS provided us with deeper insights into the power of vector-based search engines and their applications in academic research. AI and natural language processing: We gained hands-on experience with natural language processing, specifically for summarizing and generating insights from large volumes of text. Time management: Working under tight deadlines helped us improve our ability to prioritize features and manage time effectively to meet deliverables.

What's next for Vector AI

Improved AI capabilities: We aim to improve the AI’s ability to generate even more accurate and nuanced summaries and recommendations, possibly by integrating more advanced models or domain-specific data. Enhanced file parsing: Expanding the platform to handle more file types, such as DOCX or PPT, and improving the accuracy of PDF text extraction. Collaboration features: Implementing a feature where researchers can share papers, collaborate on notes, or discuss findings directly within the platform.

Built With

• pgvector
• postgresql
• python
• react
• supabase
• tailwind
• typescript
• vite