RNexus

Inspiration

I was inspired by the frustration every student, researcher, and engineer faces during literature reviews. As a researcher myself, I realize that important ideas are often buried across disciplines, contradictory results are hard to identify, and discovering genuinely new research directions can take weeks of manual reading.

I wanted to build a tool that helps people understand the research landscape and uncover connections that people might miss.

What it does

Research Nexus is an AI-powered research discovery platform.

Users can upload a paper (PDF, DOI, PMID, arXiv link, or title), and the system automatically:

• Extracts the paper’s research question, methodology, findings, limitations, and future work

• Searches the scientific literature for related work

• Builds an interactive knowledge graph of connected papers

• Identifies supporting, contradictory, and cross-disciplinary research

• Generates research gaps and unexplored opportunities

• Suggests potential collaborators and emerging trends

Challenges I ran into

1. Scientific PDFs are messy. Extracting clean text, figures, and references from different paper formats was surprisingly difficult.
2. Finding meaningful connections. Simple semantic similarity produced many irrelevant matches. I had to combine embeddings with relationship classification and citation signals.
3. Contradiction detection. Papers often disagree subtly rather than explicitly, so I designed prompts that compare claims, populations, and methodologies.
4. Visualizing large graphs. Knowledge graphs can become overwhelming quickly, so I focused on clustering, confidence scores, and interactive exploration.

Accomplishments that I'm proud of

• Built an end-to-end pipeline from PDF upload → AI analysis → interactive research map.

• Generated research-gap suggestions that can inspire future experiments.

• Delivered a demo where users can understand a research field in minutes instead of hours.

  What I learned

  Cross-disciplinary innovation often comes from recognizing shared patterns across unrelated fields. Additionally, a great AI product is not only about model quality; visualization and user experience dramatically affect insight generation.

  What's next for RNexus

• Literature review generator with fully cited outputs.

• Side-by-side paper comparison.

• Real-time alerts for newly published related work.

• Collaborative workspaces for research teams.

• Voice-guided exploration of a research field.

• Institution and lab discovery for collaboration opportunities.

Built With

• html5
• openai
• tailwindcss
• vanillajs