Insurance Library Sample: Synthetic Commercial P&C Submission Documents

RootCauseAnalytics provides a public sample from the RCA Insurance Library. It contains two complete synthetic commercial property and casualty (P&C) submission packs, comprising 14 documents across 8 document types. The dataset includes ground truth and engineered cross-document inconsistencies, with a synthetic disclaimer on every page.

Use Cases

• Train models for document classification based on the 8 commercial P&C document types mentioned.
• Develop information extraction pipelines for entities and values within synthetic insurance documents.
• Test algorithms for detecting engineered inconsistencies across multiple related documents.
• Benchmark synthetic data generation techniques for replicating complex, multi-document business processes.

Strengths

• Contains 14 complete PDF documents forming two distinct submission packs.
• Includes ground truth labels and engineered inconsistencies for validation tasks.
• Covers 8 distinct commercial P&C document types, such as broker submission emails and loss run reports.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

RootCauseAnalytics

Collection Method

Synthetically generated; no real policyholder, claimant, broker, insurer, or customer data.

Freshness

Last updated 2026-05-15 09:11:46; freshness should be verified.