By name: source-evaluation description: 来源评估 — 评估信息来源的可靠性和时效性，过滤低质量内容 version: "1.0.0" license: MIT metadata: author: blackplume233 actant-tags: "evaluation,credibility,quality,researcher" layer: auxiliary agent: researcher
Source Evaluation
Overview
Evaluate reliability and timeliness of information sources; filter low-quality content.
Source Authority Scoring
- Domain reputation: Known authoritative domains (e.g., official docs, .gov, .edu)
- Publisher track record: Historical accuracy; user ratings if available
- Expertise signals: Author credentials, institutional affiliation
Freshness Assessment
- Extract publication/update date from page metadata or content
- Score by recency; deprioritize stale content for time-sensitive topics
- Handle missing dates; infer from context when possible
Bias Detection
- Identify promotional, opinion, or advocacy content vs factual
- Flag potential conflicts of interest (e.g., vendor-authored comparisons)
- Surface bias level to user; do not auto-filter, allow user choice
Cross-Reference Verification
- Check if claims are corroborated by multiple independent sources
- Weight corroboration in overall score
Citation Tracking
- Track which sources cite which; build citation graph
- Prefer well-cited sources; flag uncited or circular citations