API Documentation

Eunomia is a framework for data governance specifically designed for LLM-based applications.

It is designed to be modular and flexible, allowing users to tailor data governance strategies to diverse use cases while maintaining precision and efficiency.

The Primitive Operations

There are two primitive operations in the framework, identification and editing. These operations are combined into a core component called an Instrument. Each Instrument is a self-contained unit designed to handle specific data governance tasks.

An Instrument consists of:

• A specific identification operation: the operation that identifies certain text entities, such as PII (e.g., names, addresses, gender) or customized entities tailored to the use case, like client names, revenue figures, or user-specific data.
• A defined editing operation: the operation that transforms the identified entities based on the requirements. The content can be redacted (context-aware or not) or transformed into aggregated or extracted insights, such as a total sum of transactions or tool usage summaries.

To see all available instruments, please refer to this section.

Data Governance in Concert

While each Instrument operates independently, effective data governance often requires addressing multiple entities and tasks simultaneously. To achieve this, multiple instruments can be combined to work together, forming what we call an Orchestra.

The Orchestra coordinates the execution of multiple instruments, allowing users to tackle complex data governance challenges.