Collections

Overview

Collections are a powerful feature in VulcanLab that allow you to organize and curate research materials into focused, topic-specific repositories. Think of a collection as a digital research folder where you gather excerpts, previous research results, and research questions related to a specific subject area.

Collections serve as the foundation for VulcanLab's advanced Deep Research feature, which can automatically synthesize all your collected materials into comprehensive, academic-quality reports.

What is a Collection?

A collection is a curated set of research materials organized around a specific topic or research question. Collections help you:

  • Organize disparate research materials in one place
  • Contextualize your research with descriptions and tags
  • Synthesize information through Deep Research
  • Track what you've read and what matters most
  • Collaborate (future feature) by sharing collections with others

Collection Metadata

Each collection has:

  • Name (required): A descriptive title for your research area
  • Description (optional): Details about the collection's scope and purpose
  • Tags (optional): Comma-separated keywords for organization and filtering
  • Timestamps: Automatic tracking of creation and modification dates
  • Item Count: Number of items in the collection

What Can You Put in a Collection?

Collections can contain three types of items:

1. Excerpts

Excerpts are highlighted passages or quotes from your library—academic papers, books, articles, or other documents you've indexed in VulcanLab.

What they contain: - The exact text passage from your source material - Source metadata (title, authors, year, work ID) - Heading breadcrumbs showing where the excerpt appears in the document - Line number references for precise location

How to add them: - While viewing search results or browsing documents - Click "Add to Collection" on any passage - Select or create a collection

Example use cases: - Key quotes supporting your thesis - Definitions of important concepts - Data points or statistics - Methodology descriptions - Contrasting viewpoints

2. Research Results

Research Results are complete answers generated by VulcanLab's RAG (Retrieval-Augmented Generation) system in response to research questions you've asked.

What they contain: - The original research question (query text) - The full generated response with citations - Metadata about which sources were used - Preview of the response content

How to add them: - From previous RAG query results - Click "Add to Collection" on any research result - Useful for building on prior work

Example use cases: - Answers to related sub-questions - Background research completed earlier - Comparative analyses you want to reference - Literature reviews on adjacent topics - Synthesized explanations to build upon

3. Research Queries

Research Queries are research questions or prompts you've formulated that relate to your collection's topic.

What they contain: - The query text (research question) - Query metadata and ID

How to add them: - Save interesting questions as you think of them - Add queries from your RAG history - Plan future research directions

Example use cases: - Questions to explore later - Sub-questions that break down complex topics - Variations of your main research question - Related areas you want to investigate

Creating and Managing Collections

Creating a New Collection

  1. From the Collections Page:
  2. Navigate to the Collections section
  3. Click "Create New Collection"
  4. Enter a name (required, max 255 characters)
  5. Add a description (optional but recommended)
  6. Add tags (optional, comma-separated)

  7. Click "Create"

  8. From the Add to Collection Modal:

  9. While adding an item from search or RAG results
  10. Click "Create New Collection" in the modal
  11. Fill in details and add the item in one step

Viewing Your Collections

Collections List View:

The main collections page shows all your collections in a table with: - Name with description tooltip on hover - Tags for quick categorization - Item Count showing how many items are in each - Created Date and Last Modified dates - Actions (delete, navigate to detail)

You can: - Sort by any column - Search by name or description - Filter by tags - Click any row to view collection details

Collection Detail View

When you open a collection, you see:

Header Section: - Collection name and metadata - Item count and last modified date - Tags (editable) - Description (click to edit inline)

In-Progress Research: - Active manual or automated research sessions - Resume buttons for paused research - Progress indicators

Deep Research Button: - Available when you have 5 or more items - Launches comprehensive research synthesis - See Deep Research Guide for details

Collection Items Table: Shows all items in your collection with: - Checkbox: Select items for bulk operations - Order: Numeric field you can edit to reorder items - Type Badge: Visual indicator (Excerpt, Research Result, Research Query) - Resource: Link to the original source with expand button - Note: Personal notes you can add inline - Metadata Card: Expandable detailed information - Actions: Delete individual items

Working with Collection Items

Adding Items

From Search Results: 1. Perform a search in VulcanLab 2. Find a relevant passage or result 3. Click "Add to Collection" 4. Choose an existing collection or create a new one 5. Optionally add a note 6. Click "Add"

From RAG Results: 1. Complete a research query 2. Review the generated results 3. Click "Add to Collection" on useful results 4. Select or create a collection 5. Add optional notes

Organizing Items

Reordering: - Click on any item's order field - Enter a new number (supports decimals for precise ordering) - Press Enter to save - Items are sorted by order value

Adding Notes: - Click on the note field for any item - Type your thoughts, context, or reminders - Notes auto-save when you click away - Use notes to: - Record why this item matters - Add context or interpretation - Note connections to other items - Flag items for follow-up

Viewing Metadata: - Click the expand button on any item - See enriched metadata specific to the item type: - Excerpts: Full source information, breadcrumbs, preview - Research Results: Original query, response preview - Research Queries: Full query text

Bulk Operations

Bulk Delete: 1. Select multiple items using checkboxes 2. Click "Delete Selected" 3. Confirm the action 4. Selected items are permanently removed

Collection Strategies and Best Practices

Focused vs. Broad Collections

Focused Collections (Recommended): - Single research question or narrow topic - 10-30 highly relevant items - Clear, specific scope - Better for Deep Research synthesis - Example: "Machine Learning Bias in Healthcare"

Broad Collections: - Multiple related sub-topics - 30-100+ items - General area of interest - Good for exploratory research - Example: "Artificial Intelligence Overview"

Recommendation: Start focused. You can always create multiple related collections rather than one massive collection.

Curation Tips

  1. Quality over Quantity:
  2. Don't add everything—be selective
  3. Each item should contribute something unique

  4. Remove items that become irrelevant

  5. Use Descriptive Names:

  6. "Climate Policy 2015-2025" > "Climate Stuff"
  7. "Neural Network Interpretability Methods" > "ML Research"

  8. Names should indicate scope clearly

  9. Write Good Descriptions:

  10. Explain the collection's purpose
  11. Note what's included and excluded

  12. Helps with future recall and Deep Research planning

  13. Tag Thoughtfully:

  14. Use consistent tag vocabulary
  15. Include: domain, methodology, time period, etc.

  16. Examples: "climate-science", "meta-analysis", "2020s", "policy"

  17. Add Meaningful Notes:

  18. Why did you include this item?
  19. How does it relate to others?
  20. What's the key takeaway?

  21. Notes improve context for Deep Research

  22. Include Diverse Sources:

  23. Mix of supporting and contrasting views
  24. Different methodologies or approaches
  25. Various publication dates
  26. Multiple authors or institutions

  27. Diversity improves research quality

  28. Maintain and Prune:

  29. Periodically review collections
  30. Remove outdated or less relevant items
  31. Add new items as you discover them
  32. Keep collections living and current

When to Create Multiple Collections

Consider splitting into separate collections when: - You have distinct sub-questions requiring different sources - One topic has 50+ items (hard to navigate) - Materials serve different purposes (literature review vs. methodology) - Time periods are distinct (historical vs. contemporary) - You want different Deep Research reports for different angles

Deep Research Integration

Collections are the foundation for VulcanLab's Deep Research feature, which automatically synthesizes your collected materials into comprehensive research reports.

Requirements

To use Deep Research, you need: - At least 5 items in your collection - A mix of item types is ideal (excerpts + research results work best) - Related materials on a cohesive topic

How It Works

Deep Research analyzes your entire collection and: 1. Understands your research topic from collection name, description, and content 2. Generates focused sub-questions that break down your topic 3. Finds existing research results to reuse (cost and time savings) 4. Selects relevant excerpts for each sub-question 5. Generates detailed, cited answers for each question 6. Evaluates quality (citation coverage, source diversity, coherence) 7. Synthesizes everything into a polished final report

Two Modes

Manual Mode: - Step-by-step control - Review and approve at each stage - Edit research plan and questions - Choose what to reuse - Full transparency

Automated Mode: - One-click operation - Handles everything automatically - Progress indicators - Notification when complete - Great for quick insights

For complete details, see the Deep Research Collections Guide.

Use Cases and Examples

Academic Research

Collection: "Algorithmic Bias in Criminal Justice" - 12 excerpts from peer-reviewed papers on predictive policing - 6 excerpts from case studies of biased systems - 4 research results on fairness metrics in ML - 2 research queries about policy interventions

Outcome: Deep Research generates a 15-page literature review with 25+ citations, ready for thesis chapter.

Technical Documentation

Collection: "Kubernetes Security Best Practices" - 20 excerpts from official K8s security docs - 8 excerpts from CVE reports and postmortems - 5 research results on container isolation methods - 3 research queries about zero-trust architectures

Outcome: Comprehensive security guide for internal engineering team.

Product Research

Collection: "User Authentication UX Patterns" - 15 excerpts from UX research papers - 10 excerpts from competitor product analyses - 7 research results on passwordless auth methods - 4 research queries about accessibility concerns

Outcome: Product requirements document with evidence-based recommendations.

Learning and Study

Collection: "Quantum Computing Fundamentals" - 18 excerpts from textbooks on quantum mechanics - 12 excerpts from quantum algorithm papers - 8 research results explaining key concepts - 5 research queries for self-testing

Outcome: Comprehensive study guide with clear explanations and exam prep material.

Technical Details

Data Model

Collections are stored with: - Unique ID (auto-generated) - User ownership (linked to your account) - Metadata (name, description, tags) - Timestamps (created, updated) - Relationships to items and research sessions

Collection items are stored with: - Unique ID - Collection ID (foreign key) - Item type (enum: excerpt, research_result, research_query) - Link (formatted path to original resource) - Note (optional text) - Order (decimal number for flexible sorting) - Timestamps (added, modified)

Items use specific link formats to reference original resources:

Excerpts:

/search/result/{work_id}/{start_line}/{end_line}

Example: /search/result/123/45/67

Research Results:

/rag/{query_id}/results/{result_id}

Example: /rag/456/results/789

Research Queries:

/rag/{query_id}

Example: /rag/456

API Endpoints

Collections are accessible via REST API:

Method Endpoint Purpose
GET /api/v1/collections/ List all collections (paginated)
POST /api/v1/collections/ Create new collection
GET /api/v1/collections/{id} Get collection details
PATCH /api/v1/collections/{id} Update collection metadata
DELETE /api/v1/collections/{id} Delete collection
POST /api/v1/collections/{id}/items Add item to collection
PATCH /api/v1/collections/{id}/items/{item_id} Update item
DELETE /api/v1/collections/{id}/items/{item_id} Delete item
DELETE /api/v1/collections/{id}/items Bulk delete items
GET /api/v1/collections/{id}/items/{item_id}/metadata Get enriched metadata

Frequently Asked Questions

Q: How many items should a collection have?

A: There's no strict limit, but 10-30 items is ideal for focused research. You need at least 5 items to use Deep Research.

Q: Can I move items between collections?

A: Currently, you'd need to add the item to the new collection and remove it from the old one. A direct move feature may be added in the future.

Q: What happens if I delete a source document?

A: The collection item remains, but the link may become invalid. Excerpts store a preview, so you'll still have the text even if the source is removed.

Q: Can I export a collection?

A: You can export Deep Research reports generated from collections. Direct collection export may be added in the future.

Q: How are collections different from bookmarks or folders?

A: Collections are intelligent—they store not just links, but rich metadata and context. They integrate with Deep Research for automatic synthesis. They're research tools, not just organizational tools.

Q: Can I share collections with others?

A: Not currently, but collection sharing is planned for a future release.

Q: Do collections have a size limit?

A: No hard limit, but very large collections (100+ items) may be slower to process in Deep Research.

Q: What if I have duplicate items?

A: Deep Research deduplicates content automatically, so duplicates won't appear multiple times in reports.

Next Steps

  • Create your first collection on a topic you're researching
  • Add 5-10 relevant items from your search results
  • Try Deep Research to see automatic synthesis in action
  • Read the Deep Research Collections Guide for advanced workflows

Collections transform VulcanLab from a search tool into a comprehensive research assistant. Start curating, and let Deep Research do the synthesis work for you.