Knowledge Graph & Causal Discovery

What is Knowledge Graph & Causal Discovery?

Knowledge Graph & Causal Discovery is an MCP server that performs causal inference over multi-domain research data. It orchestrates 17 Apify actors across five source domains—academic, biomedical, regulatory, economic, and safety—and applies ten peer-reviewed causal algorithms to discover directed causal structure, estimate treatment effects, and reason about counterfactuals. It is designed for researchers, data scientists, and AI agents that need to go beyond correlation.

How to use Knowledge Graph & Causal Discovery?

Add the server URL and your Apify bearer token to your MCP client configuration (e.g., Claude Desktop, Cursor, Windsurf). The server exposes eight tools such as discover_causal_structure, compute_interventional_effects, and simulate_counterfactuals. Each tool call returns structured JSON with mathematical scores and supporting evidence.

Key features of Knowledge Graph & Causal Discovery

• Always-live data fetched fresh from source APIs per call.
• Parallel execution of up to 17 actors per query.
• Standby mode eliminates cold-start latency.
• Pay-per-call pricing ($0.035–$0.050 per tool).
• MCP-native—works with Claude Desktop, Cursor, Windsurf, Cline.
• Covers 17 data sources across five domains.
• Applies 10 peer-reviewed causal algorithms.
• Returns structured JSON with scores and evidence.

Use cases of Knowledge Graph & Causal Discovery

• Drug safety signal detection: combine PubMed, ClinicalTrials.gov, and FDA adverse event reports to discover causal edges between compounds and adverse outcomes.
• Policy impact assessment: estimate causal effects of regulatory interventions on economic outcomes using FRED and World Bank data.
• Systematic review and evidence synthesis: scan thousands of papers simultaneously and classify causal claims by strength and evidence level.
• Counterfactual reasoning for legal and regulatory causation: compute Probability of Necessity and Sufficiency via twin network method.
• Knowledge graph completion in biomedical AI: generate RotatE embeddings for link prediction in sparse graphs.
• Data acquisition prioritization: use Shapley values to quantify each data source's marginal contribution to causal graph quality.

FAQ from Knowledge Graph & Causal Discovery

What data sources are available?

The server accesses 17 data sources across five domains: academic (OpenAlex, Semantic Scholar, Crossref, arXiv, CORE), biomedical (PubMed, ClinicalTrials.gov, OpenFDA, NIH Reporter), regulatory (Federal Register, Congress.gov, Data.gov), economic (FRED, World Bank), and safety (CPSC, CFPB, Wikipedia). A detailed table is provided in the README.

What causal inference algorithms are supported?

Ten peer-reviewed algorithms are implemented across eight tools, including FCI skeleton learning, GES with BIC scoring, Pearl's do-calculus with the ID algorithm, twin network counterfactuals, TMLE estimation, RotatE knowledge graph embeddings, sheaf cohomology consistency checking, and Shapley source attribution.

How is Knowledge Graph & Causal Discovery accessed and authenticated?

The server is a remote MCP endpoint accessed via JSON-RPC over HTTP. Authentication is via a Bearer token (your Apify API token) passed in the request header. The URL is https://knowledge-graph-causal-discovery-mcp.apify.actor/mcp.

What is the pricing model?

Each tool call costs between $0.035 and $0.050. An 8-tool pipeline (one call per tool) costs under $0.35. There are no monthly subscriptions; pricing is strictly per-call.

Does the server use cached or live data?

Data is always-live. Every tool call fetches fresh results from the source APIs; no stale snapshots or cached indexes are used.