WHY CHOOSE LUMINA
Advanced AI-powered screening that adapts to your research criteria
SEMANTIC SEARCH
SPECTER embeddings trained on 100M+ scientific papers understand the meaning of research text, not just keywords. Papers are ranked by semantic similarity to your inclusion criteria.
ACTIVE LEARNING
The system learns from every decision you make. After each screening choice, a machine learning model retrains and re-ranks remaining papers to surface relevant ones faster.
AI ASSISTANT
Unsure about a paper? Ask Pixel-Bot for an intelligent recommendation with detailed reasoning based on your inclusion criteria. Get a second opinion when you need it.
DUAL SCORING
Every paper receives two AI scores: a semantic similarity score from embeddings and a classifier confidence score from machine learning. Combined, they prioritize the most relevant papers.
WEB-BASED
No installation required. No Python knowledge needed. Just sign up, upload your papers, and start screening in minutes from any browser.
FLEXIBLE IMPORT/EXPORT
Import papers from any database via RIS or CSV format. Export your screened results to CSV, RIS, or Excel for use in your systematic review.
EU DATA RESIDENCY
All servers hosted in the European Union. Your research data stays in the EU, fully compliant with GDPR and European data protection regulations.
THREE-LAYER AI APPROACH
Semantic Understanding
SPECTER embeddings encode the deep meaning of scientific abstracts, matching papers to your criteria conceptually.
Machine Learning
TF-IDF classifier learns your inclusion patterns and improves with every decision you make.
AI Assistance
Pixel-Bot provides on-demand explanations and recommendations for difficult screening decisions.
VALIDATED PERFORMANCE
Tested on real systematic review datasets with published ground truth labels from the SYNERGY benchmark.
FREQUENTLY ASKED QUESTIONS
How does Lumina help with systematic reviews?
Lumina uses AI to prioritize papers by relevance to your inclusion criteria. Instead of screening papers in random order, you screen the most likely relevant papers first—finding most inclusions while reviewing fewer papers.
What makes Lumina's AI different?
Lumina combines semantic understanding (SPECTER embeddings) with active learning (ML model retraining after every decision) and AI assistance. This three-layer approach delivers better paper prioritization than single-method tools.
Do I need to install anything?
No! Lumina is entirely web-based. Just create an account and start screening—no software to install, no Python required.
What file formats are supported?
Import papers via RIS or CSV format from any database (PubMed, Scopus, Web of Science, etc.). Export your results to CSV, RIS, or Excel.
Where is my data stored?
All Lumina servers are hosted in the European Union. Your research data never leaves the EU, ensuring full GDPR compliance and meeting European data protection requirements.
READY TO TRY LUMINA?
Join researchers who screen papers faster with AI assistance.