value_patterns.json

{
  "version": "3.0.0",
  "description": "价值信号识别模式库 + 置信度启发式打分",

  "patterns": {
    "replyToUnlock": {
      "description": "回复解锁类信号",
      "priority": "high",
      "action": "auto_reply",
      "regex": [
        "回复.{0,10}可见",
        "回复.{0,10}下载",
        "回复.{0,10}查看",
        "reply.{0,10}unlock",
        "reply.{0,10}download",
        "reply.{0,10}see",
        "reply.{0,10}view",
        "comment.{0,10}see",
        "留言.{0,10}查看"
      ]
    },

    "hiddenContent": {
      "description": "隐藏内容类信号",
      "priority": "high",
      "action": "login_required",
      "regex": [
        "登录后查看",
        "登录可见",
        "登录后下载",
        "VIP.{0,10}可见",
        "VIP.{0,10}下载",
        "hidden content",
        "login to view",
        "login required",
        "members only"
      ]
    },

    "downloadLink": {
      "description": "下载链接类信号",
      "priority": "critical",
      "action": "download",
      "regex": [
        "mega\\.nz/[^\"]+",
        "mediafire\\.com/[^\"]+",
        "drive\\.google\\.com/[^\"]+",
        "pan\\.baidu\\.com/[^\"]+",
        "megaup\\.net/[^\"]+",
        "rapidgator\\.net/[^\"]+",
        "dropbox\\.com/[^\"]+",
        "1fichier\\.com/[^\"]+",
        "zippyshare\\.com/[^\"]+",
        "uploadgig\\.com/[^\"]+",
        "nitroflare\\.com/[^\"]+",
        "turbo\\.to/[^\"]+"
      ]
    },

    "extractCode": {
      "description": "提取码类信号",
      "priority": "critical",
      "action": "save_code",
      "regex": [
        "密码[：:][\\s]*([a-zA-Z0-9]{4,8})",
        "提取码[：:][\\s]*([a-zA-Z0-9]{4})",
        "pass[：:][\\s]*([a-zA-Z0-9]+)",
        "password[：:][\\s]*([a-zA-Z0-9]+)",
        "pwd[：:][\\s]*([a-zA-Z0-9]+)",
        "解压密码[：:][\\s]*([a-zA-Z0-9]+)",
        "密码是[：:][\\s]*([a-zA-Z0-9]+)"
      ]
    },

    "attachment": {
      "description": "附件文件类信号",
      "priority": "high",
      "action": "download_attachment",
      "extensions": [
        ".zip", ".rar", ".7z", ".tar", ".gz",
        ".pdf", ".doc", ".docx", ".xls", ".xlsx",
        ".py", ".js", ".ts", ".sh", ".bat",
        ".exe", ".msi", ".dmg", ".apk",
        ".torrent", ".magnet"
      ]
    },

    "taskGate": {
      "description": "任务门槛类信号",
      "priority": "medium",
      "action": "log_requirement",
      "regex": [
        "关注公众号",
        "分享到",
        "点赞.{0,10}可见",
        "转发.{0,10}解锁",
        "关注.{0,10}下载",
        "share.{0,10}unlock",
        "like.{0,10}download"
      ]
    }
  },

  "valueDensity": {
    "description": "价值密度评估因子",
    "factors": {
      "hasDownloadLink": 30,
      "hasCode": 20,
      "hasAttachment": 25,
      "replyCount": 1,
      "likeCount": 0.5,
      "authorReputation": 15,
      "freshness": 10
    },
    "threshold": 50
  },

  "credibilityScoring": {
    "description": "Credibility heuristic — turns the 'trustworthy' characteristic from an adjective into a computable score. A claim backed by concrete data (numbers, prices, case counts) is worth more than an opinion. Use this to rank extracted knowledge by evidence strength before writing it into any report.",
    "signals": {
      "hasConcreteNumbers": {
        "weight": 25,
        "description": "Specific figures: success rates, traffic counts, ranking positions, timeframes",
        "regex": ["\\d+%", "\\$\\d+", "\\d+\\s*(days?|weeks?|months?)", "\\d+x", "\\d{2,}\\s*(backlinks?|keywords?|posts?|accounts?)"]
      },
      "hasPricing": {
        "weight": 20,
        "description": "Real cost data signals first-hand experience, not theory",
        "regex": ["\\$\\d+", "\\d+\\s*(usd|dollars?)", "per\\s*(month|article|link|click)", "/mo", "/month"]
      },
      "hasCaseEvidence": {
        "weight": 25,
        "description": "Concrete case references: 'I ran X for Y and got Z'",
        "regex": ["case study", "my (site|account|client)", "i (tested|ran|tried)", "results?:", "before.{0,20}after", "实测", "亲测", "案例"]
      },
      "hasTimeframe": {
        "weight": 10,
        "description": "Recent date anchors signal freshness and verifiability",
        "regex": ["202[5-9]", "this (year|month|week)", "recently", "last (month|week)"]
      },
      "opinionOnly": {
        "weight": -15,
        "description": "Penalty: pure opinion with no evidence — 'I think', 'probably', 'should work'",
        "regex": ["i think", "probably", "should work", "in my opinion", "maybe", "i guess", "我觉得", "应该可以"]
      }
    },
    "tiers": {
      "high": "score >= 50 — evidence-backed, safe to surface as primary finding",
      "medium": "score 20-49 — partial evidence, surface with caveat",
      "low": "score < 20 — opinion-grade, log but do not feature"
    },
    "note": "This is a public, generic evidence-strength heuristic. Production-grade source verification (provenance tagging, freshness cross-checks, contradiction detection) is a deeper discipline — see the V3 Roadmap in README for the community track."
  }
}