Glossary
Glossary of data licensing terms
Plain definitions of the terms that come up when a company licenses its SOPs, documents and code for AI training.
A
- Agent training data
- Agent training data is data used to teach or test AI agents that carry out multi-step tasks with software tools, such as recorded task sequences, the policies and procedures an agent must follow, and the end states that show whether a task was done correctly.
- Anonymization
- Anonymization is the processing of data so that no person can be identified from it by any means reasonably likely to be used, whether by the holder or by anyone else.
C
- Code review history
- Code review history is the record of how proposed code changes were reviewed before being merged, including the diffs, reviewers' line comments, requested changes, replies and approval decisions, usually kept in a pull request or merge request system.
- Commit history
- Commit history is the version-control record of every change made to a codebase, in which each commit stores the change itself, a message describing it, the author and committer and their timestamps, linked in order back to the first commit.
- Copyleft license obligations
- Copyleft license obligations are the conditions in licences such as the GNU GPL that require anyone who distributes the code, or modified versions of it, to pass on the same freedoms, source code and licence terms.
D
- Data broker vs data partner
- A data broker collects and sells personal information about people it has no direct relationship with, while a data partner is a company that licenses data from its own operations under a negotiated agreement.
- Data licensing agreement
- A data licensing agreement is a contract in which a data owner grants another party defined rights to use a copy of its data, for agreed purposes and an agreed period, usually in exchange for payment, without selling the data itself.
- Data provenance
- Data provenance is the documented record of where a dataset came from, who created it, under what rights and terms it was collected, and how it has been changed since.
- Data retention clause
- A data retention clause is the part of a contract that says how long a party may keep data it receives, when and how it must delete or return it, and what it may keep afterwards.
- Data sample
- A data sample is a small, representative extract of a dataset that a prospective buyer reviews to judge its content, format and quality before agreeing to license the full set.
- Datasheet for datasets
- A datasheet for datasets is a structured document that describes a dataset's purpose, contents, collection process, recommended uses and limits, so a recipient can judge whether it fits their use.
- De-identification
- De-identification is the process of removing or transforming the details in a dataset that identify, or could reasonably identify, a person, so the records can be used without revealing who they are about.
E
- Enterprise data monetization
- Enterprise data monetization is a company earning measurable value from the data it holds, either indirectly by using it to improve its own operations and products or directly by licensing or selling it to others.
- Exclusive license
- An exclusive license gives one licensee the sole right to use data or other property within an agreed scope and period, so the owner cannot license it to anyone else within that scope.
- Expert reasoning traces
- Expert reasoning traces are records of the intermediate steps a skilled person took to reach a decision or solve a problem, such as the evidence checked, the options weighed and the reasons given, rather than only the final answer.
F
- Field of use
- A field of use is the specific purpose, product area or market in which a licensee may use licensed data or other property, set out in the license so that every other use stays with the owner.
K
- Knowledge base data
- Knowledge base data is the content of a company's internal or customer-facing knowledge base, such as wiki pages, runbooks, help articles and FAQs, together with their edit history, used as training or retrieval data for AI models.
L
- License term
- A license term is the period during which a licensee may use licensed data or other property, set by a start date, an end date or event, and any renewal or termination rights.
N
- Non-exclusive license
- A non-exclusive license gives a licensee the right to use data or other property while the owner stays free to use it and to license the same thing to others.
P
- PII redaction
- PII redaction is the removal or masking of personally identifiable information, such as names, contact details and account numbers, from documents, records or code before they are shared.
- Post-training data
- Post-training data is the data used to refine an AI model after its initial pretraining, such as example requests paired with ideal responses for supervised fine-tuning, rankings or preferences between responses, and tasks with checkable outcomes.
- Process mining data
- Process mining data is an event log exported from a company's business systems, in which each record ties an activity to a case, usually with a timestamp and the person or system that performed it, so the real sequence of work can be reconstructed.
- Pseudonymization
- Pseudonymization is the replacement of names and other direct identifiers in a dataset with codes or aliases, while the key that links them back to real people is kept separately.
R
- Re-licensing
- Re-licensing is granting a new license to data that already exists, such as records a company already holds in its software or a dataset it has licensed before, instead of creating new data for the deal.
S
- SOP data
- SOP data is a company's standard operating procedures, work instructions and checklists, together with the revision history and execution records that show how those procedures are actually followed, used to train or evaluate AI models.
fearofgod.ai
Turn your company’s data into long-term revenue.
We license private and niche data from companies like yours, de-identify it and pay you for it. Partnerships typically range from $50K to $3M+. Your company can be paid more than once, through renewals and repeat licences.