We use cookies to improve your experience and analyse site traffic.

Platform

Data Schema Analysis

Know your data governance posture before you write a line of model code. Data Schema Analysis operates at the metadata level by design — your production data stays in your infrastructure. We analyse structure, not content. Automated classification across 17+ schema types against GDPR data categories and EU AI Act Article 10 requirements.

17+ Schema Types

Comprehensive data architecture coverage

From relational databases to vector stores, event streams to ML model registries — schema analysis covers the full modern data stack.

Relational (SQL)

PostgreSQL, MySQL, SQL Server, Oracle — table structures, column types, foreign keys, constraints

Document Store

MongoDB, Couchbase, DynamoDB — document schemas, nested structures, polymorphic collections

Graph Database

Neo4j, Neptune, ArangoDB — node types, relationship schemas, property graphs

Vector Store

Pinecone, Weaviate, Milvus, pgvector — embedding dimensions, metadata schemas, index configurations

Time-Series

InfluxDB, TimescaleDB, Prometheus — metric schemas, tag structures, retention policies

Data Lake (Parquet/Delta)

Delta Lake, Apache Iceberg, Parquet — column schemas, partition structures, evolution history

Object Storage

S3, GCS, Azure Blob — file type classification, metadata extraction, naming convention analysis

Key-Value Store

Redis, DynamoDB, etcd — value structure inference, TTL patterns, access patterns

Columnar (OLAP)

BigQuery, Snowflake, Redshift — warehouse schemas, materialised views, data lineage

Event Stream

Kafka, Kinesis, Pulsar — event schemas (Avro, Protobuf, JSON Schema), topic structures

Search Index

Elasticsearch, OpenSearch, Solr — mapping definitions, analyser configurations, field types

File System

CSV, JSON, XML, YAML — structural inference, header detection, nested object analysis

API Schema

OpenAPI/Swagger, GraphQL SDL, gRPC Protobuf — endpoint schemas, request/response types

Data Catalog

Apache Atlas, DataHub, Amundsen — metadata schemas, lineage graphs, classification taxonomies

Feature Store

Feast, Tecton, Hopsworks — feature schemas, entity definitions, transformation lineage

ML Model Registry

MLflow, Weights & Biases, Neptune — model metadata, training configuration, dataset references

Data Warehouse (dbt)

dbt models, sources, snapshots — schema definitions, column descriptions, test specifications

GDPR Classification

Automated data category detection

Schema analysis identifies fields that are likely to contain regulated data categories — flagging them for governance review before any data processing begins.

Personal Data

Any information relating to an identified or identifiable natural person — names, email addresses, IP addresses, location data, online identifiers. The baseline classification that triggers GDPR obligations.

Special Category Data

Data revealing racial or ethnic origin, political opinions, religious beliefs, trade union membership, genetic data, biometric data, health data, sex life, or sexual orientation. Processing generally prohibited unless explicit conditions are met.

Criminal Convictions Data

Personal data relating to criminal convictions and offences. Processing restricted to official authority supervision or specific authorised conditions. Common in background check and fraud detection systems.

Pseudonymous Data

Personal data of persons under 16 (or lower age set by member state, minimum 13). Requires parental consent for information society services. Enhanced protections apply throughout the data lifecycle.

EU AI Act Article 10

Training data governance requirements

Article 10 establishes specific requirements for the data used to train, validate, and test high-risk AI systems. Schema analysis addresses these obligations at the structural level.

Training Data Quality

Training, validation, and testing datasets must be subject to appropriate data governance — including data collection, relevant preparation, labelling, cleaning, enrichment, and aggregation. Schema analysis identifies data quality indicators before training begins.

Representativeness

Datasets must be relevant, sufficiently representative, and to the best extent possible, free of errors and complete. Schema analysis flags coverage gaps, demographic imbalances, and geographic underrepresentation in the data structure.

Bias Indicators

Examination of possible biases that are likely to affect the health and safety of persons, have a negative impact on fundamental rights, or lead to discrimination. Schema analysis identifies proxy variables and correlation patterns at the structural level.

Data Lineage

The origin, scope, characteristics, and availability of training data must be documented. Schema analysis maps data sources, transformation chains, and provenance metadata across your data architecture.