1. Determine whether it constitutes an advertisement:
   • With direct or indirect promotion purpose → constitutes an advertisement
   • For claims of "non-advertisement", verify three exemption conditions (own media + objective introduction + no promotion). If any condition is not met, it will be treated as an advertisement
   • If it is judged as non-advertisement, output the conclusion directly and terminate the review
2. Identify industry categories (multiple selections allowed):

   • real estate

     food

     alcohol

     health food

     food for special medical purposes

     infant formula food

   • medical care

     pharmaceuticals

     medical devices

     pesticides

     veterinary drugs

   • education and training

     investment and investment attraction

   • internet advertising

     live streaming

   • public service advertisement

   • general goods/services

     (not belonging to any of the above special categories)
3. Identify advertising forms (multiple selections allowed):

   • information flow

     soft article/review/experience

     pop-up/splash screen

     bidding ranking

     live stream e-commerce

   • outdoor advertising

     traditional media

     self-media/official account

4. Identify whether it involves:

   • endorsement/recommendation

     promotion/prize-attached sales

     trademark/patent publicity

按风险等级从高到低排列

• 违规类型： [政治红线/虚假误导/绝对化用语/形式违规/行业专项违规/...]
• 风险等级： [高/中/低]
• 问题描述： [具体描述广告中的违规内容]
• 原文摘录： "[摘录广告原文中的违规表述]"
• 违反条款： [对应的法规条款或审核规则]
• 修改建议： [具体的修改方向或替代表述]

1. Write the review results into a JSON file with the following structure:

{
  "basicInfo": {
    "广告素材": "素材描述",
    "审核日期": "2026-02-11",
    "广告定性": "构成广告",
    "行业分类": "一般商品/服务",
    "广告形式": "店铺促销告示"
  },
  "conclusion": "通过 | 需修改后通过 | 不予通过",
  "riskLevel": "高 | 中 | 低",
  "issues": [
    {
      "title": "问题简述",
      "type": "违规类型",
      "risk": "高 | 中 | 低",
      "description": "问题描述",
      "quote": "原文摘录",
      "violation": "违反条款",
      "suggestion": "修改建议"
    }
  ],
  "suggestions": "合规建议（多段落用 \\n 分隔）",
  "disclaimer": "免责声明（可选，有默认值）"
}

1. Run the generation script:

node scripts/generate_report.js <input.json> [output.docx]

• High Risk: Political red line (one-vote veto), false advertising, missing statutory administrative license, prohibited advertising
• Medium Risk: Absolute terms, non-compliant form (no "advertisement" mark), missing statutory mandatory items, standard clauses infringing consumer rights
• Low Risk: Vague expression but not constituting false content, unclear citation source, incomplete detail marking

1. Strict handling for doubtful cases: Tend to mark risks for vague expressions rather than release them
2. Layered output: High-risk issues must be prominently marked, low-risk issues can be explained in combination
3. Provide alternative solutions: Try to provide operable modification suggestions for each problem
4. Distinguish factual judgment from legal judgment: Matters that cannot be judged only by advertising text (such as whether qualification documents are complete) are marked as "suggested verification" rather than directly judged as illegal