Karlsen Randi, Borrás Morell Jose Enrique, Fernández Luque Luis, Traver Salcedo Vicente
Department of Computer Science, University of Tromsø, Norway.
Stud Health Technol Inform. 2013;192:1008.
Health information retrieval and YouTube can be used as powerful tools to improve user's health knowledge. However, YouTube videos must be carefully analysed in order to avoid misleading, inaccurate, obsolete and incorrect health content. We present an approach for re-ranking health videos obtained from YouTube, called Domain-based ranking. Our system automatically identifies videos coming from trusted sources (channels), such as hospitals and health organizations, and re-ranks YouTube results so that such videos are presented first in the ranking list. Video and channel metadata are used to automatically determine if a video is provided by a trusted source. The approach is tested and results show that the amount of relevant and reliable videos ranked within top-10 increase when using Domain-based ranking, compared with the original YouTube ranking.
健康信息检索和YouTube可作为增强用户健康知识的有力工具。然而，必须仔细分析YouTube视频，以避免误导性、不准确、过时和错误的健康内容。我们提出了一种对从YouTube获取的健康视频重新排名的方法，称为基于领域的排名。我们的系统会自动识别来自可信来源（频道）的视频，如医院和健康组织，并对YouTube搜索结果重新排名，以便这些视频在排名列表中首先呈现。视频和频道元数据用于自动确定视频是否由可信来源提供。该方法经过测试，结果表明，与原始的YouTube排名相比，使用基于领域的排名时，排名前10的相关且可靠的视频数量有所增加。
