Grading the Narrators: An Isnād–Rijāl Framework for Claim-Level Provenance in Multi-Agent Knowledge Systems
Year: 2026
Type: preprint
Abstract: Modern multi-agent knowledge systems increasingly accumulate knowledge through chains of autonomous transformations rather than direct retrieval. Existing provenance work records what happened — execution traces, tool calls, evidence links — but does not generally provide a unified framework to grade who transformed a claim and therefore how much the result should be trusted. Classical Islamic hadith science confronted a structurally similar problem: deciding whether knowledge transmitted through chains of human narrators should be accepted. Over centuries of scholarly refinement, it developed a rigorous methodology — isnād (a complete transmission chain attached to every claim), rijāl (systematic grading of each narrator’s integrity and precision), weakest-link chain evaluation, corroboration upgrades through independent chains, and matn criticism (content evaluated independently of chain quality). This paper transfers that methodology to AI system design. The framework proposes: (1) a formal mapping from hadith-science concepts to multi-agent knowledge pipelines; (2) a concrete relational schema implementing claim chains and a graded narrator registry; (3) a decision matrix combining chain grade with content criticism; and (4) a preliminary case study in which a prototype self-maintaining knowledge base surfaced 19 genuine cross-framework contradictions in undergraduate physics texts, demonstrating the content-criticism substrate the framework requires. The framework targets factual knowledge accumulation, not open-ended or creative generation. The framework is intentionally modular: different deployments may instantiate different grading functions while preserving the same architectural principles. This is a systems-architecture and epistemology paper: it proposes a framework, a relational schema, and a validation of the framework’s content-criticism substrate; empirical validation of end-to-end narrator grading is deliberately scoped as future work and the paper is explicit about which of its claims that evidence does and does not yet support.
Source: arXiv (Cornell University)
Author: Raja, Ali Zahid
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7167492914
Cites:
Cited by:
Related to: