• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Workshops on Internet Science >Methodological Challenges for Detecting Interethnic Hostility on Social Media
【24h】

Methodological Challenges for Detecting Interethnic Hostility on Social Media

机译：检测社交媒体中整群敌意的方法论挑战

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Detection of ethnic hate speech and other types of ethnicity representation is user texts is an important goal both for social and computer science, as well as for public policy making. To date, quite a few algorithms have been trained to detect hate speech, however, what policy makers and social scientists need are complete pipelines, from definition of ethnicity to a user-friendly monitoring system able to aggregate results of large-scale social media analysis. In this essay, the author summarizes the experience of development of such a system in a series of projects under the author's leadership. All steps of the offered methodology are described and critically reviewed, and a special attention is paid to the strengths and the limitations of different approaches that were and can be applied along the developed pipeline. All conclusions are based on prior experiments with several large datasets from Russian language social media, including 15 000 marked up texts extracted from a representative one-year collection of 2.7 million user messages containing ethnonyms.
机译：种族仇恨言论的检测和其他类型的种族代表性是用户文本是社会和计算机科学以及公共政策制定的重要目标。迄今为止，已经训练了相当多的算法以检测仇恨言论，但是，哪些政策制定者和社会科学家需要完整的管道，从种族的定义到用户友好的监控系统，能够聚合大规模社交媒体分析的结果。在本文中，提交人总结了在作者领导下的一系列项目中开发此类系统的经验。所提供的方法的所有步骤都被描述并批评，并对不同方法的优点和局限性进行了特别关注，并且可以沿着发达的管道应用。所有结论都是基于先前的实验，其中来自俄语社交媒体的几个大型数据集，其中包括从包含民族名义的270万用户消息的代表性一年收集提取了15 000个标记的文本。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号