MediaGuard

MediaGuard detects and prevents digital advertising fraud. MediaGuard and FraudSensor work together to detect invalid traffic (also known as IVT) and ensure that your ads are seen by real humans—not bots in disguise. If you’re a HUMAN user, our Snowflake integration lets you access the full scope of data collected by our tools, including the pre-bid IVT predictions made by MediaGuard and the post-bid IVT detection and analysis performed by FraudSensor. These datasets offer event-level data about individual bid requests and impressions, including HUMAN’s “bot or not” decisions, detailed threat breakdowns, and information about the traffic sources driving each event. Sample Tables: - pre-bid IVT prediction data (MediaGuard): mediaguard_prebid - sampled pre-bid IVT prediction data (MediaGuard): mediaguard_prebid_sampled - post-bid IVT detection and analysis (FraudSensor): mediaguard_postbid Fields Included in MediaGuard pre-bid data: - supplier_id: The supplier id for the requests - publisher_id: The publisher id for the requests - device_type: The device type for the requests - media_type: The media type for the requests - ivt_bot_category: The various IVT categories detected - requests: Approximate number of requests and 16 more fields. Fields Included in MediaGuard sampled pre-bid data: - supplier_id: The supplier id for the requests - publisher_id: The publisher id for the requests - device_type: The device type for the requests - media_type: The media type for the requests - domain: The domain for the requests - app_id: The app id for the requests - ivt_bot_category: The various IVT categories detected - requests: Approximate number of requests and 16 more fields. Fields Included in MediaGuard post-bid: - ct: Query parameters used to fetch HUMAN's payload excluding c1 to c10 - custom: c1 to c10 query parameters used to fetch HUMAN's payload - app_mismatch: Whether the app id claimed by the user and detected by HUMAN matched or not (mobile only) - detected_app_id: The app id detected by HUMAN (mobile only) - environment: The user's environment used for the connection - ivt_bot_category_str: The IVT categories detected - domains: Various domains related to the ad viewed by the user - bot: Whether ad was viewed by a bot or not - givt: Whether the bot was GIVT or not - sivt: Whether the bot was SIVT or not and 21 more fieldsMediaGuard可检测并防范数字广告欺诈行为。MediaGuard与FraudSensor协同工作，可检测无效流量（Invalid Traffic，下文简称IVT），确保广告仅被真实人类用户浏览，而非伪装的机器人。 若您为人类用户，我们的Snowflake集成方案可让您获取工具采集的全量数据，包括MediaGuard生成的竞价前IVT预测数据，以及FraudSensor完成的竞价后IVT检测与分析数据。此类数据集提供单条竞价请求与广告展示的事件级数据，涵盖HUMAN的「机器人/非机器人」判定结果、详细威胁分类明细，以及驱动各事件的流量来源信息。 示例数据表： - 竞价前IVT预测数据集（MediaGuard）：mediaguard_prebid - 采样版竞价前IVT预测数据集（MediaGuard）：mediaguard_prebid_sampled - 竞价后IVT检测与分析数据集（FraudSensor）：mediaguard_postbid MediaGuard竞价前数据集包含以下字段： - supplier_id：请求对应的供应商标识符 - publisher_id：请求对应的发布商标识符 - device_type：请求对应的设备类型 - media_type：请求对应的媒体类型 - ivt_bot_category：检测到的各类IVT类别 - requests：请求的近似数量 另有16个附加字段。 MediaGuard采样版竞价前数据集包含以下字段： - supplier_id：请求对应的供应商标识符 - publisher_id：请求对应的发布商标识符 - device_type：请求对应的设备类型 - media_type：请求对应的媒体类型 - domain：请求对应的域名 - app_id：请求对应的应用标识符 - ivt_bot_category：检测到的各类IVT类别 - requests：请求的近似数量 另有16个附加字段。 MediaGuard竞价后数据集包含以下字段： - ct：用于获取HUMAN负载的查询参数（不含c1至c10） - custom：用于获取HUMAN负载的c1至c10查询参数 - app_mismatch：用户声明的应用标识符与HUMAN检测结果是否匹配（仅适用于移动场景） - detected_app_id：HUMAN检测到的应用标识符（仅适用于移动场景） - environment：用户建立连接时使用的环境 - ivt_bot_category_str：检测到的IVT类别 - domains：与用户浏览的广告相关的各类域名 - bot：广告是否被机器人浏览 - givt：该机器人是否为GIVT - sivt：该机器人是否为SIVT 另有21个附加字段。

搜集汇总

背景与挑战

背景概述

MediaGuard数据集包含预投标和后投标两个阶段的数字广告欺诈检测数据，通过与FraudSensor协作识别无效流量(IVT)。该数据集提供事件级数据，包括供应商/发布者信息、设备类型、IVT分类等字段，帮助区分真实用户与机器人流量。