A concept graph that people govern
Themes and concepts connect to verses through an ontology. Taaweel treats that graph as something to be curated: machine-proposed links go through automated checks and human adjudication before they are trusted.
What it does
The ontology labels verses and relates concepts to each other. Language models can propose new edges and labels, but proposals are not accepted by default. They enter a queue.
Reviewers can approve, reject, remap or merge. Alongside that human step, a policy engine checks structure: no self-loops, cycle checks, subtree constraints, and minimum confidence and provenance criteria.
Governance needs a policy
The tooling supports review, but what to approve is a scholarly decision. Institutions using the ontology should define who may adjudicate and what is uploaded to external AI platforms.
- Run and worker logs make labelling traceable.
- Admin tools exist for controlled data reset, to be wrapped in an organisation's own procedures.
- Role-based access separates ordinary users from administrators.
Step by step
Propose
Models suggest labels and relations between concepts.
Check
Automated QA rejects structurally invalid proposals.
Adjudicate
Reviewers approve, reject, remap or merge.
Explore
Navigate the graph and open the verses behind each node.
Controls
Graph explorer
Move through concepts and open the labelled verses behind a node.
Policy checks
No self-loops, cycle detection, subtree rules, confidence and provenance thresholds.
Human adjudication
Approve, reject, remap and merge operations with a record of who did what.
Worker queue
Background labelling runs with scheduling, batch limits and retry handling.
Common questions
Is the ontology complete?
No. It grows through labelling runs and review, and is designed to be extended under governance.
Can I export the graph?
Graph data lives in the application database and is available through its API for authorised administrators.
Who can change the ontology?
Administrators and designated reviewers under role-based access.
See Taaweel run on your question
Tell us the question you are researching. We will walk through a run and what the evidence ledger and audit look like.