Asoba Zorora Documentation
Home Insights Contact Us

Academic Research Use Case

Use Zorora’s deep research capabilities for comprehensive academic research and literature reviews.

Scenario

You’re writing a research paper on “Recent advances in transformer architectures for natural language processing” and need to:

Academic Research Example Screenshot
(Placeholder - Add screenshot showing academic research results with citations and credibility scores)

Step-by-Step Guide

Step 1: Start Research Query

Terminal:

zorora
[1] ⚙ > What are the latest developments in transformer architectures for natural language processing?

Web UI:

  1. Open http://localhost:5000
  2. Enter query: “latest developments in transformer architectures for natural language processing”
  3. Select depth: Thorough (for comprehensive research)
  4. Click “Start Research”

API:

from engine.research_engine import ResearchEngine

engine = ResearchEngine()
state = engine.deep_research(
    "latest developments in transformer architectures for natural language processing",
    depth=3  # Thorough depth for academic research
)

Step 2: Review Sources

Academic Sources Found:

Source Credibility:

Review Sources:

for source in state.sources_checked:
    if source.source_type == "academic":
        print(f"{source.title}: {source.credibility_score:.2f}")
        print(f"  URL: {source.url}")

Step 3: Explore Citation Trails

Citation Following:

Access Citation Graph:

# Citation relationships are stored in research state
# Access via research data
research_data = engine.load_research(state.started_at.strftime("%Y%m%d_%H%M%S"))
citation_graph = research_data.get('citation_graph', {})

Step 4: Cross-Reference Claims

Grouped Claims:

Review Findings:

for finding in state.findings:
    print(f"Claim: {finding.claim}")
    print(f"Agreement: {finding.agreement_count} sources")
    print(f"Sources: {finding.sources}")

Step 5: Synthesize Findings

Synthesis Includes:

Access Synthesis:

print(state.synthesis)

Step 6: Save Research

Save for Later:

research_id = engine.save_research(state)
print(f"Research saved: {research_id}")

Access Later:

research_data = engine.load_research(research_id)

Example Output

Synthesis

Based on recent research, transformer architectures for natural language processing have seen significant developments in several areas:

**Efficiency Improvements:**
- Sparse attention mechanisms [Academic: arXiv:2024.12345]
- Model compression techniques [Academic: arXiv:2024.67890]
- Efficient training methods [Web: example.com/article]

**Architecture Innovations:**
- Multi-head attention variants [Academic: arXiv:2024.11111]
- Position encoding improvements [Academic: arXiv:2024.22222]

**Key Consensus:**
- Sparse attention is widely accepted as effective [8 sources agree]
- Model compression maintains quality [6 sources agree]

**Areas of Disagreement:**
- Best position encoding method [3 sources disagree]

Sources

High Credibility Sources:

Medium Credibility Sources:

Best Practices

Query Formulation

Be Specific:

Include Context:

Use Research Keywords:

Depth Selection

Quick (depth=1):

Balanced (depth=2):

Thorough (depth=3):

Source Evaluation

Check Credibility:

Follow Citations:

Consider Context:

Advanced Usage

Multiple Research Queries

Related Topics:

[1] ⚙ > What are the latest developments in transformer architectures?
[2] ⚙ > How do attention mechanisms work in transformers?
[3] ⚙ > What are efficient training methods for transformers?

Compare Results:

Citation Graph Analysis

Explore Relationships:

# Load research with citation graph
research_data = engine.load_research(research_id)
citation_graph = research_data.get('citation_graph', {})

# Identify key papers (most cited)
key_papers = identify_key_papers(citation_graph)

Cross-Reference Verification

Verify Claims:

# Find claims with high agreement
high_agreement = [f for f in state.findings if f.agreement_count >= 5]

# Find claims with disagreement
disagreement = [f for f in state.findings if f.agreement_count <= 2]

Troubleshooting

No Academic Sources Found

Problem: Research returns no academic sources

Solution:

Low Credibility Scores

Problem: All sources have low credibility

Solution:

Citation Graph Empty

Problem: Citation graph not populated

Solution:

See Also