Epistemic Typing as a PostgreSQL Table Access Method: Adversarial Conflict Resolution Under Confidence Forgery and Sybil Coordination
KNDB, a PostgreSQL 18 access method, resists confidence forgery and Sybil writes using engine-assigned epistemic types.
KNDB is a PostgreSQL 18 table access method that labels each row MEASURED, INFERRED, or DERIVED and resolves write-time conflicts inside heap callbacks. On confidence-forgery workloads it beat a confidence-only baseline by 63 points on Book-Author and 92.7 points on Zheng; neutralizing the kind lattice removed the gain. Versus TruthFinder, CRH, CATD, and ACCU it is competitive or dominant once source density saturates, but loses below saturation on Zheng and scores zero on last-writer-wins temporal benchmarks.
- PostgreSQL 18 access method types every row epistemically.
- Beats confidence-only scoring by 63 and 92.7 points.
- Ablating the kind lattice eliminates the measured gain.
- CRH and ACCU win below saturation on Zheng.
- Scores zero where ground truth is last-writer-wins.
- Products
- PostgreSQLKNDB
From the original article
1 min read at arxiv.orgWe describe KNDB, a PostgreSQL 18 table access method (TAM) that types every row with an engine-assigned epistemic kind (MEASURED, INFERRED, or DERIVED) and resolves per-slot conflicts inside every write-time heapam callback.
By Venkata Maguluri · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.