fair-protocol/docs/moderation/tools/setup-ozone.md

# Setting Up and Running a FAIR-Compatible Ozone Labeler Instance

| <!-- --> | <!-- -->   |
|----------|------------|
| Status   | Proposal   |
| Date     | 2025-07-22 |

This guide provides an overview for developers and organizations interested in setting up and operating their own independent Ozone-based Labeler Service that is compatible with the FAIR ecosystem. Running your own labeler allows you to contribute specialized moderation signals, cater to specific community needs, or offer alternative perspectives on trust and safety.

Before proceeding, ensure you are familiar with:
- The core concepts of label-based moderation in FAIR, as detailed in [Ozone Labeling System](../../ozone-labeling-system.md).
- FAIR's governance principles, particularly those related to [Moderation Services](../../governance/moderation-services.md).
- Basic principles of Decentralized Identifiers (DIDs) and the AT Protocol, upon which Ozone is built.

**Disclaimer:** This guide provides FAIR-specific guidelines. For the core setup and operation of an Ozone instance, refer to the official Ozone documentation from the Bluesky project: [https://github.com/bluesky-social/ozone](https://github.com/bluesky-social/ozone). This document assumes a technical understanding of deploying and managing web services.

**Future Notes:** In the event that a large number of people or organizations wish to operate a labeling service, FAIR may create a specific package to aid installation and configuration.

## 1. Purpose and Planning

Before deploying a labeler, clearly define:

* **Your Moderation Scope:** What types of content or entities will you focus on? (e.g., security of packages, WCAG accessibility, compliance with a specific regional policies such as GDPR, PIPEDA, or other legislation, or factual information for cataloguing packages such as license, software features, or genres of other content.)
* **Your Label Set:** What specific labels will you issue? (e.g., `myorg:security-audited`, `community:focus-accessibility`, `vendor:official-partner`, `wcag:2.2AA`, etc.) Define these clearly.
* **Your Moderation Criteria:** What are the transparent rules and processes by which you will apply these labels?
* **Your Target Audience:** Who do you expect to consume your labels? (e.g., a specific community, developers focused on a certain type of package)

## 2. Core Ozone Instance Setup

Setting up the base Ozone service involves several technical steps, which should be guided by the official Ozone documentation. At a high level, this typically includes:

* **Deployment Environment:** Choosing and configuring a server environment (e.g., cloud virtual machine, containerized setup).
* **Ozone Service Installation:** Deploying the Ozone application software.
* **Database Configuration:** Setting up and connecting a compatible database (e.g., PostgreSQL).
* **Service Configuration:** Configuring Ozone with its operational parameters, such as its domain, connection to the AT Protocol network (if applicable for identity resolution or publishing), etc.
* **Identity for the Labeler Service (DID):**
    * Your Labeler Service itself will need a Decentralized Identifier (DID). This DID will be associated with the labels it issues, ensuring accountability and allowing other services to identify the source of the labels.
    * Follow AT Protocol/Ozone procedures for creating or assigning a DID to your service instance.

**Action:** Refer to the latest official Ozone installation and configuration guides for detailed, up-to-date instructions.

## 3. Configuring for FAIR Compatibility

To ensure your Ozone labeler instance can effectively participate in the FAIR ecosystem:

* **Understanding FAIR Entity Identification:**
    * FAIR Repositories, Aggregators, and packages will be identified by DIDs or other standardized unique identifiers within the FAIR network.
    * Familiarize yourself with how FAIR plans to structure its "Package/Repository Registry" and any specific ATProto record types FAIR might define for these entities (see [Ozone Labeling System](../../ozone-labeling-system.md)). Your labeler will need to target these identifiers.

* **Defining and Publishing Your Labels:**
    * **Schema (Recommended):** While not strictly enforced by FAIR for third-party labelers, consider defining a clear, public schema for the labels your service will issue. This helps consumers understand the meaning and intent behind your labels (e.g., `com.myorg.labels.securityReviewLevel` with possible values `basic`, `thorough`).
    * **Public Documentation:** Publicly document the labels your service issues, their meanings, and your criteria for applying them. This builds trust and encourages adoption.

* **Making Labels Discoverable:**
    * Ensure your Ozone instance's API endpoint for querying labels is publicly accessible.
    * Consider how your labels will be discoverable within the broader AT Protocol ecosystem or any FAIR-specific indexing services.

## 4. Operational Best Practices

* **Transparency:**
    * Clearly state who operates the labeler service.
    * Publish your moderation policies and criteria.
    * If you offer a process for disputing labels issued by your service, document it clearly.
* **Reliability and Maintenance:** Ensure your service is reliably hosted and maintained. Downtime can impact services that subscribe to your labels. You may want to consider high-availability architecture and the use of CDNs.
* **Security:** Secure your Ozone instance and its administrative interfaces to prevent unauthorized access or label manipulation.
* **Communication:** Provide a clear point of contact for your labeler service.

## 5. Announcing Your Labeler Service to the FAIR Community

Once your labeler is operational and you have documented its purpose and policies:

* Inform the FAIR community through appropriate channels (e.g., FAIR forums, developer mailing lists, or *specific channels which may be defined by FAIR in the future*).
* Provide details on:
    * The DID of your labeler service.
    * The API endpoint to query labels.
    * A link to your documentation explaining your labels and policies.

## 6. Interaction with the FAIR Ecosystem

* **Subscription by Aggregators or Clients:** FAIR Aggregators (including FAIR's own), and client applications (package browsers or installers), and end users can choose to subscribe to labels from any discoverable labeler service, including third-party or private ones.
* **No Guaranteed Endorsement:** Running a FAIR-compatible labeler _does not_ imply an official endorsement or mandatory subscription by FAIR or its official services. Each subscribing service or user will decide which labelers to trust and follow.
* **Community Trust:** Building a reputation for issuing accurate, consistent, and useful labels is key to gaining adoption and trust within the FAIR community.

---

This document provides a high-level guide. The specific technical steps for Ozone deployment are subject to the official Ozone project documentation and may evolve. The FAIR community may also develop further recommendations or tools to aid in the setup and discovery of FAIR-compatible labelers.