contradiction_detect

You detect cross-source contradictions by pairing claims from different sources and analyzing their propositions. The Skill outputs a structured list of disputes featuring type, severity, source_a, source_b, and supporting evidence for literature reviews and technical reports.

Cross-source claim pairing for risk assessment compares factual claims from multiple sources to surface inconsistencies. It characterizes explicit and implicit contradictions using a typology and severity labels to enable better quality control and decision support.

Yes, this tool works with investigative datasets to find factual disputes. It applies to cross-source claim sets in investigative datasets to surface disputes, ensuring claim-pairs are limited to distinct sources for accurate contradiction detection.

To detect methodological and interpretive contradictions, provide a list of claims annotated with their sources. The Skill analyzes these cross-source claim sets, applies typology labels like factual, methodological, or interpretive, and outputs a structured contradiction report.

The best way to surface inconsistencies across technical reports is to use cross-source claim pairing and comparison. This approach identifies explicit and implicit contradictions and outputs structured signals for downstream resolution, prioritization, and human-in-the-loop escalation.

Yes, you can use this for data quality control in large datasets. It systematically characterizes contradictions across multiple information sources, outputting structured signals that support downstream resolution, prioritization, and human-in-the-loop escalation.