农产品类快手直播带货违法监测预警数据
收藏资源简介：
对采集的快手平台带货品类为农产品类的达人直播视频内容进行转译分析，对达人口播语言内容进行处理、分析，根据达人在直播过程中对预先设置的违规敏感词（比如：100%纯天然，绝对无农药残留，特效治疗，绿色有机，预防疾病，抗癌，抗衰老，政府推荐，无激素，快速生长剂，口感绝佳，品质卓越等）违反的次数和频率，依据触发条件规则提出警告或处理。为诸暨市市场监督局管理区域内规范企业快手直播行为，提供数据支持。将采集完成的直播视频进行进行预处理，第一步：基于原始视频文件，以最大10分钟单位对原始视频进行切片。第二步：对于已完成的切片视频，进行视频内容转语音操作。第三步：对于已完成视频转语音操作的切片，进行语音转文本操作。第四步：使用OCR技术对原始视频中抓取的图片进行文字提取操作。第五步：将所得到的文字内容与违法预警关键词库进行匹配。最终运用多标准决策分析模型，对主播在直播过程中出现的违规语句进行分析计算，得出违法预警值和是否预警判断。 违法预警值=（违法预警单关键词命中次数*0.25）+（违法预警组合关键词命中次数* 0.3）+（图片识别命中预警组合关键词个数*0.35）+（直播间近一个月历史违规记录数*0.1） 通过公式计算出最终违法预警值，违法预警值 ≤1 时，不触发预警提示，违法预警值 ＞1 时触发违法预警提示。
We perform transcription and analysis on the collected live stream videos of agricultural product-focused live-streaming influencers on the Kuaishou platform, and process and analyze the oral language content of these influencers. Warnings or penalties will be issued in accordance with the trigger condition rules, based on the frequency and count of violations of pre-configured sensitive prohibited keywords (such as: 100% pure and natural, absolutely no pesticide residues, specific curative effect, green organic, disease prevention, anti-cancer, anti-aging, government recommendation, hormone-free, rapid growth agents, excellent taste, superior quality, etc.) during the live streams. This dataset provides data support for standardizing the live streaming behaviors of enterprises within the jurisdiction of the Zhuji Municipal Market Supervision Administration via the Kuaishou platform. The collected live stream videos undergo the following preprocessing steps: 1. Segment the original video files into clips with a maximum duration of 10 minutes each; 2. Convert the content of the segmented videos into speech; 3. Convert the speech extracted from the segmented videos into text; 4. Extract text from images captured from the original videos using Optical Character Recognition (OCR) technology; 5. Match the obtained text content against the prohibited early warning keyword database. Finally, a multi-criteria decision analysis model is employed to analyze and calculate the violation statements made by the influencers during their live streams, to derive the Violation Early Warning Value and the early warning judgment result. The Violation Early Warning Value is calculated via the following formula: Violation Early Warning Value = (Number of hits for single prohibited early warning keywords * 0.25) + (Number of hits for combined prohibited early warning keywords * 0.3) + (Number of combined early warning keywords identified via image recognition * 0.35) + (Number of historical violation records of the live broadcast room in the past month * 0.1) The final Violation Early Warning Value is calculated using the above formula. If the value is ≤ 1, no early warning prompt will be triggered; if the value is > 1, a violation early warning prompt will be issued.
