textTrue

TextTrue is a conceptual framework and accompanying markup system designed to embed truth-valued metadata in natural language texts. It provides a standardized way to annotate factual claims with veracity scores and provenance links, enabling automated verification and more transparent attribution.

The concept emerged in the late 2020s as part of efforts to combat misinformation and support trustworthy content. Proponents describe TextTrue as a lightweight layer layered on top of existing document formats, interoperable across platforms via a defined vocabulary and an open data model. The project is maintained by a collaborative community rather than a single company.

Core elements include sentence-level truth tags, a numeric confidence score between 0 and 1, and references to evidence sources. Each claim may also include a cryptographic hash for integrity, a timestamp, and a pointer to supporting documents. The data model is designed to be compatible with RDF-like graphs and can be exported to JSON-LD or other serializations.

In practice, authors annotate claims when writing articles or reports. Verification engines gather corroborating evidence from trusted databases and publish a revised score alongside the text. Readers can filter or sort content by truth score, examine cited sources, and assess the reliability of individual claims.

Limitations include dependence on available evidence and potential biases in source selection. Privacy concerns and the risk of over-reliance on automated scores are noted. TextTrue is intended as a supplement to, not a replacement for, rigorous editorial and fact-checking processes.