Documentation
- Query Refinement
- Research Lenses
- AI-Powered Semantic Search
- Unbiased Computational Analysis
- Research Landscape
- Research Landscape Synthesis
- Trends and Momentum
- AI Research Assistant
- Citation-Grounded Answers
- Hypothesis Generation
- Methodology Generation
- Network Analysis
- Foundational Papers
- Hub Papers
- Bridge Papers
- Novel Leads
- PDF Export
Documentation
AI Research Assistant
Chat with your evidence base to get instant, synthesized answers. Ask complex follow-up questions and get answers from full-text analysis. Every claim is footnoted to the source papers for full traceability.
How It Works
The AI Research Assistant is powered by Google Gemini and works with your selected papers:
- Select Papers: Choose papers from your search results that you want to analyze
- Ask Questions: Pose questions about the selected papers using natural language
- Get Answers: Receive synthesized answers grounded in the paper content, with citations to source papers
- Follow Up: Ask follow-up questions that build on previous answers, maintaining context throughout the conversation
Types of Questions You Can Ask
The AI Research Assistant can help with various types of analysis:
- Summarization: "Summarize the key findings from these papers"
- Comparison: "Compare the methodologies used across these studies"
- Specific Details: "What were the specific p-values reported for treatment efficacy?"
- Mechanisms: "Explain how the signaling pathway works according to these papers"
- Contradictions: "Are there any contradictory findings among these studies?"
- Gaps: "What research gaps do these papers identify?"
Model Options
BioSkepsis uses Google Gemini models for answering questions:
- Gemini 2.5 Flash Preview (default): Fast and efficient, ideal for most queries
- Gemini 2.5 Pro: More advanced reasoning capabilities for complex analysis
Full Text Analysis
For deeper analysis, the system can access full text of papers:
- Automatic Download: Full text is automatically downloaded for papers that meet your filter criteria (similarity score, impact) when needed for follow-up questions
- Comprehensive Context: Full text provides access to methods, results, discussion, and conclusions, not just abstracts
- Efficient Resource Usage: Full text is only downloaded when needed, not for every paper
- Persistent Storage: Once downloaded, full text is reused in subsequent questions
Citation-Grounded Answers
Every answer from the AI Research Assistant includes citations:
- Source Attribution: Claims are linked back to specific papers
- Verification: You can verify every statement by checking the cited papers
- Traceability: Full traceability from synthesis back to source publications
- No Hallucinations: Answers are grounded only in your selected papers, eliminating AI fabrications
Best Practices
To get the most from the AI Research Assistant:
- Select relevant papers: Choose papers that are directly related to your question
- Be specific: More specific questions yield more focused answers
- Build on context: Use follow-up questions to explore topics in depth
- Verify citations: Always check cited papers to verify important claims
- Use filters: Apply filters to focus on high-quality or recent papers when needed