隐私关注视野下移动社交平台用户隐私保护协议规制模型构建及实证研究  

The Construction and Empirical Study of the Regulation Model of Privacy Protection Agreement for Social Platform Users from the Perspective of Privacy Concern

作　　者：代沁泉 王斌清 周静虹 Dai Qinquan

机构地区：[1]武汉大学信息管理学院,湖北武汉430072 [2]武汉大学图书情报国家级实验教学示范中心,湖北武汉430072 [3]山西大学图书馆,山西太原030006

出　　处：《情报理论与实践》2023年第10期77-85,共9页Information Studies:Theory & Application

基　　金：国家社会科学基金青年项目“算法分发环境下隐性数字弱势群体信息公平实现路径研究”的成果之一,项目编号:22CTQ038。

摘　　要：[目的/意义]构建社交平台用户隐私协议文本规制评估模型框架,规范社交平台用户隐私保护协议及提升协议规制评估效率。[方法/过程]首先,结合用户隐私关注研究、用户隐私保护相关法规文本、社交平台用户隐私保护规制研究进行规制指标提取,并在指标融合的基础上构建规制评估框架模型;其次,基于规制框架模型,对获取的社交平台隐私协议进行分类模型训练数据的人工标注;最后,通过对部分文本数据训练,获得分类效果最好的模型参数,进而对收集到的33个社交平台用户隐私保护协议文本进行分类及规制评估。[结果/结论]社交平台用户隐私协议的规制结果显示,其存在内容缺失、内容粗糙、权责不明在内的多层问题,基于此从政策监督主体、平台、用户维度出发细化完善保护策略。[局限]人工编码和筛选有一定的主观性,选取的社交平台类型多而总量不大。[Purpose/significance]To build a framework of regulation evaluation model for social platform user privacy agreement text,to standardize social platform user privacy protection agreement and improve the efficiency of agreement regulation evaluation.[Method/process]Firstly,this paper combines user privacy concern research,user privacy protection related regulation text,and social platform user privacy protection regulation research to extract regulation indicators,and constructs a regulation evaluation framework model based on the integration of indicators;secondly,based on the regulation framework model,the obtained social platform privacy protocols are manually labeled with the training data of the classification model;and then the model parameters with the best classification effect are obtained by training some text data,and then the collected 33 social platform user privacy protection protocols are classified and evaluated for regulation.[Result/conclusion]The results of the regulation of social platform user privacy agreements show that there are multiple layers of problems,including missing content,rough content,and unclear rights and responsibilities,based on which the protection strategy is refined and improved from the dimensions of policy supervision subject,platform,and user.[Limitations]Manual coding and screening are somewhat subjective,with many social platform types selected but not a large total.

关 键 词：社交平台 隐私关注 隐私保护协议 规制评价 BiLSTM-Attention模型 

分 类 号：G206[文化科学—传播学] G252