Video Brand Safety & Suitability Scores

Qortex provides brand safety and suitability scores for online video assets. Each video is scored on a 0-5 scale across 11 brand safety categories including violence, hate speech, adult content, drugs, weapons, crime, profanity, alcohol/tobacco, gambling, political extremism, and negative news.

<p><br/></p>

Additionally, videos receive an overall brand suitability assessment (None/Low/Medium/High) with human-readable explanations of why content was flagged, plus the specific GARM (Global Alliance for Responsible Media) categories triggered.

<p><br/></p>

This dataset enables:

- Pre-bid brand safety filtering for programmatic advertising

- Post-bid verification and reporting

- Content moderation signal enrichment

- Publisher inventory quality assessment

- Domain-level safety profiling

<p><br/></p>

Each video is linked to the page URL and domain where it was found, enabling inventory-level safety decisioning. Updated regularly as new videos are analyzed.

原始信息汇总

数据集概述：Video Brand Safety & Suitability Scores

提供商：Qortex

简要描述：该数据集为在线视频内容提供品牌安全性与适宜性评分，覆盖11个风险类别，并包含整体品牌适宜性评估。

---

核心功能与价值

• 品牌安全评分：对每个视频在11个类别（暴力、仇恨言论、成人内容、毒品、武器、犯罪、脏话、酒精/烟草、赌博、政治极端主义、负面新闻）上进行0-5分的打分。
• 品牌适宜性评估：提供整体评估等级（无/低/中/高），并附有人类可读的标记原因解释，以及触发的特定GARM（全球负责任媒体联盟）类别。
• 数据应用场景：
  • 程序化广告的预出价品牌安全过滤
  • 出价后验证与报告
  • 内容审核信号增强
  • 发布商库存质量评估
  • 域名级别的安全画像分析

数据字典

该数据集包含一个核心表 VIDEO_RISK_PROFILE，其关键字段如下：

字段名	数据类型	说明
LOGICAL_VIDEO_ID	字符串 (Varchar)	视频的唯一逻辑ID
ADULT_CONTENT_SCORE	数字 (Number)	成人内容风险评分 (0-5)
ADULT_CONTENT_REASON	字符串 (Varchar)	成人内容评分的人类可读解释
VIOLENCE_SCORE	数字 (Number)	暴力风险评分 (0-5)
VIOLENCE_REASON	字符串 (Varchar)	暴力评分的人类可读解释
HATE_SPEECH_SCORE	数字 (Number)	仇恨言论风险评分 (0-5)
HATE_SPEECH_REASON	字符串 (Varchar)	仇恨言论评分的人类可读解释
DRUGS_SCORE	数字 (Number)	毒品风险评分 (0-5)
DRUGS_REASON	字符串 (Varchar)	毒品评分的人类可读解释
WEAPONS_SCORE	数字 (Number)	武器风险评分 (0-5)
WEAPONS_REASON	字符串 (Varchar)	武器评分的人类可读解释
CRIME_SCORE	数字 (Number)	犯罪风险评分 (0-5)
CRIME_REASON	字符串 (Varchar)	犯罪评分的人类可读解释
PROFANITY_SCORE	数字 (Number)	脏话风险评分 (0-5)
PROFANITY_REASON	字符串 (Varchar)	脏话评分的人类可读解释
ALCOHOL_TOBACCO_SCORE	数字 (Number)	酒精/烟草风险评分 (0-5)
ALCOHOL_TOBACCO_REASON	字符串 (Varchar)	酒精/烟草评分的人类可读解释
GAMBLING_SCORE	数字 (Number)	赌博风险评分 (0-5)
GAMBLING_REASON	字符串 (Varchar)	赌博评分的人类可读解释
POLITICAL_EXTREMISM_SCORE	数字 (Number)	政治极端主义风险评分 (0-5)
POLITICAL_EXTREMISM_REASON	字符串 (Varchar)	政治极端主义评分的人类可读解释
NEGATIVE_NEWS_SCORE	数字 (Number)	负面新闻风险评分 (0-5)
NEGATIVE_NEWS_REASON	字符串 (Varchar)	负面新闻评分的人类可读解释

注：数据集中还包含VIDEO_BRAND_SUITABILITY和VIDEO_BRAND_SAFETY_SUMMARY表，提供视频适宜性等级和与页面URL、域名的关联信息。

使用示例（SQL查询片段）

1. 查找品牌安全视频：查询所有风险类别评分均为0的视频。 sql SELECT LOGICAL_VIDEO_ID, ADULT_CONTENT_SCORE, VIOLENCE_SCORE, HATE_SPEECH_SCORE FROM BRAND_SAFETY.VIDEO_RISK_PROFILE WHERE ADULT_CONTENT_SCORE = 0 AND VIOLENCE_SCORE = 0 AND HATE_SPEECH_SCORE = 0 AND DRUGS_SCORE = 0 AND WEAPONS_SCORE = 0 AND CRIME_SCORE = 0 AND PROFANITY_SCORE = 0 LIMIT 100;

2. 查看适宜性等级分布：统计不同适宜性等级的视频数量及占比。 sql SELECT SUITABILITY_LEVEL, COUNT() AS VIDEO_COUNT, ROUND(COUNT() * 100.0 / SUM(COUNT(*)) OVER(), 2) AS PCT FROM BRAND_SAFETY.VIDEO_BRAND_SUITABILITY GROUP BY SUITABILITY_LEVEL ORDER BY VIDEO_COUNT DESC;

3. 识别高风险视频：查找暴力、仇恨言论或犯罪评分较高（>=3）的视频及其原因。 sql SELECT LOGICAL_VIDEO_ID, VIOLENCE_SCORE, VIOLENCE_REASON, HATE_SPEECH_SCORE, HATE_SPEECH_REASON, CRIME_SCORE, CRIME_REASON FROM BRAND_SAFETY.VIDEO_RISK_PROFILE WHERE VIOLENCE_SCORE >= 3 OR HATE_SPEECH_SCORE >= 3 OR CRIME_SCORE >= 3 LIMIT 50;

4. 为家庭友好型广告活动筛选库存：查询适宜性等级为"无"且成人、暴力、脏话评分为0的视频及其关联的域名和页面URL。 sql SELECT LOGICAL_VIDEO_ID, DOMAIN, PAGE_URL, SUITABILITY_LEVEL FROM BRAND_SAFETY.VIDEO_BRAND_SAFETY_SUMMARY WHERE SUITABILITY_LEVEL = None AND ADULT_CONTENT_SCORE = 0 AND VIOLENCE_SCORE = 0 AND PROFANITY_SCORE = 0 LIMIT 1000;

数据更新与交付

• 更新频率：每日（Daily）更新。
• 交付方式：Secure share（安全分享）。