互联网应用页面标签库与域名归类数据服务
收藏资源简介：
1.产品核心依据与目标该产品依据运营商的上网日志入库标准（DPI）挖掘用户的上网行为特征，通过深度挖掘与分析，建立互联网上网日志特征与标签之间的精确对应关系，进而构建通用的域名识别模型，且该模型不针对某个用户。2.入参与出参入参：需监测的 APP 名称或网站地址等；出参：域名识别模型分拣出的、与输入 APP 或网站地址相关的域名列表。3. 产品处理流程四必得基于 HTTP/HTTPS 等互联网用户常用的超文本协议（其中，HTTP 指不加密的上网链接，易识别；加密的 HTTPS 链接，运营商无 URL 详情仅能获取域名，产品基于域名组合做意图预判），运用先进的机器学习技术，再结合运营商提供的 DPI 话单库对规则特征进行准确性校验（四必得不接触用户个体数据），最终建立精确的域名识别模型。4. 应用示例与合规说明例如，针对 “点击进入携程 APP” 的行为，可依据特征规则 “downlo.ctr.com” 与之对应，并利用运营商提供的 DPI 话单库（对应合同：广州四必得科技有限公司与 ** 分公司大数据业务合作协议）进行校验。产品仅提供域名使用的归类和标签，不涉及获取 APP 的隐藏信息，也不保存任何 APP 的数据。
1. Product Core Basis and Objectives This product mines users' Internet browsing behavior features in accordance with the operator's Internet access log storage standard (Deep Packet Inspection, DPI). Through in-depth mining and analysis, it establishes an accurate correspondence between Internet access log features and corresponding labels, and then constructs a general-purpose domain name recognition model that does not target individual users. 2. Input and Output Parameters Input Parameters: Names of target apps or website addresses, etc. Output Parameters: A list of domain names related to the input app or website address, sorted out by the domain name recognition model. 3. Product Processing Flow Sibide utilizes hypertext protocols commonly used by Internet users such as HTTP/HTTPS (where HTTP refers to unencrypted browsing links that are easy to identify; for encrypted HTTPS links, operators can only obtain domain names without detailed URL information, and user intentions are predicted based on domain name combinations), adopts advanced machine learning technology, and combines with the DPI call detail record database provided by the operator to verify the accuracy of rule features (Sibide does not access any individual user data), and finally establishes an accurate domain name recognition model. 4. Application Examples and Compliance Instructions For example, for the behavior of "clicking to enter the Ctrip APP", it can be matched with the feature rule "downlo.ctr.com", and verified using the DPI call detail record database provided by the operator (corresponding contract: Cooperation Agreement on Big Data Business between Guangzhou Sibide Technology Co., Ltd. and ** Branch). The product only provides classification and labeling of domain name usage, does not involve obtaining hidden information of apps, nor does it store any app data.
