source reliability
9 Articles
Conflicting Sources Fact Checking: What Explanations Must Show
A 2026 explainer on conflicting sources in fact checking: evidence-relation matrix, span removal tests and six scores, drawing on CLUE, ConfRAG (57.2% of 1,814 questions) and NatConfQA.
Privacy-Preserving Truth Discovery: A Utility Audit
A 2026 research audit of privacy-preserving truth discovery: protected units, local privacy budgets, source-weight reversals, secure aggregation, and a reproducible utility evaluation card.
LLM conflicting sources: a counterfactual fusion test
A 2026 evidence review of LLM source conflict resolution, reconstructing tabular-fusion F1 results and specifying entity masking, source-dependence tests, abstention metrics and reproducible provenance.
Source selection in data fusion: evaluating marginal evidence
A 2026 evaluation protocol for source selection in data fusion: compare held-out subsets, separate coverage from verified accuracy, test source dependence, and audit integration costs and provenance.
Truth existence in data fusion: when all sources err
A 2026 research guide to no-truth questions in data fusion, TEM source-quality assumptions, SF2013 F1 denominators, and a held-out test for candidate omission and correct abstention.
Testing Streaming Truth Discovery Under Drift
A 2026 research guide to streaming truth discovery, separating one-pass computation, evolving truth and source reliability drift with decision matrices and failure-injection metrics.
Comparing Truth Discovery Methods Without Gold Labels
A 2026 protocol for evaluating truth discovery without ground truth, comparing synthetic stress tests, sparse adjudication, observation likelihood, stability, and uncertainty.
Multiple Sources of Truth: An Assumption Audit
A 2026 assumption audit for multi-truth data integration, comparing LTM, MBM, and DART with worked claim matrices, omission semantics, provenance, and evaluation metrics.
Truth Discovery with Copying Sources: Benchmark Guide
A 2026 assumptions-first guide to truth discovery benchmarks, comparing 12-method studies, copying detection above 95% in one setup, sparse labels, metrics, and method selection.