Data Usage Policy

This policy describes how behavioral data submitted to BarkMind — including case reports, annotations, expert resolutions, and consensus records — is stored, used, and eventually released as part of the open behavioral dataset.

BarkMind exists to build the world's most structured open canine behavioral dataset. All behavioral data submitted to the platform serves this purpose. By submitting data, you are contributing to a shared scientific resource.

Submission: Cases, annotations, and resolutions are stored in our PostgreSQL database with full attribution metadata (submitter ID, timestamp, confidence level, credential status at time of submission).

Annotation period: During the open annotation period, data is visible to all platform users and accessible via the API.

Expert review: Cases receive formal expert review and consensus records. These become part of the permanent case record.

Evidence lock: When a case is resolved and evidence-locked, an immutable snapshot of all case data is created. This snapshot is what enters the dataset.

Dataset release: Locked cases are periodically bundled into dataset releases (v0.x during beta, v1.0 for the first public release). Releases are published under CC BY 4.0.

Dataset releases include:

• Case descriptions, behavioral context, and setting metadata
• All annotations with type, body, confidence level, and author attribution
• Behavioral taxonomy references for each annotation
• Expert resolution verdicts and reasoning
• Consensus records and individual expert opinions
• Contributor attribution table (username, credential status, contribution type)

Dataset releases do not include: account email addresses, private account metadata, IP addresses, or any personally identifying information beyond usernames.

Media files (images and video) are not included in bulk dataset downloads by default due to size. See the Media Rights Policy for details on media access and licensing.

The BarkMind API provides programmatic access to platform data. Public endpoints are available without authentication. Authenticated endpoints are available to registered users. Rate limiting applies to prevent abuse.

Academic or institutional research partnerships may receive enhanced dataset access under data sharing agreements that preserve contributor attribution and usage restrictions. Contact us to discuss research access.

We do not sell behavioral data to commercial entities. We do not train AI models on the data without providing attribution to contributors. We do not share individual user data (beyond what appears in dataset releases) with third parties.

The long-term goal of BarkMind includes using the dataset to train and evaluate behavioral AI systems. When this occurs, contributors whose data was used will be acknowledged in any associated publications. No AI training on this data will occur without the explicit dataset release happening first.

Data questions: [email protected]