# Sample Datasets

# Sample Datasets

Sample datasets are a practical way to evaluate Curiosity Workspace and to teach teams the core patterns:

• ingest data into a graph schema
• configure search and facets
• enable embeddings and AI-assisted workflows
• build endpoints and interfaces

# What makes a good sample dataset

A good sample dataset has:

• Multiple entity types (at least 3–5 node types)
• Relationships that support navigation and filtering (edges)
• Text fields suitable for search (titles, summaries, descriptions)
• A time dimension (timestamps) to test recency and time filters
• Enough volume to see ranking and performance behavior

# Recommended sample dataset categories

• Support and case management
  • tickets/cases, products, customers, teams, statuses
• Compliance and audit
  • policies, controls, evidence, owners, exceptions
• Engineering knowledge base
  • docs, code artifacts, services, incidents, runbooks
• Research
  • papers, authors, topics, citations, institutions

# How to use sample datasets in your docs/testing

Use sample datasets to validate:

• Graph navigation: do the relationships enable the workflows you want?
• Search relevance: can users find the right objects by keywords?
• Semantic recall: can vector search find “similar meaning” content?
• Facet usefulness: do the chosen facets match how users refine results?

# Next steps

• Implement ingestion for a dataset: Connectors
• Make it searchable: Search → Text Search