Wikipedia Culture Gap: Quantifying Content Imbalances Across 40 Language Editions
Wikipedia Culture Gap: Quantifying Content Imbalances Across 40 Language Editions
复制标题
DOI：
10.3389/fphy.2018.00054
复制
发表时间：
2018-06-06
影响因子：
3.1
通讯作者：
Laniado, David
中科院分区：
文献类型：
--
作者：
Miquel-Ribe, Marc;Laniado, David
在线百科全书维基百科是通过来自全球各地的合作努力创建的最大的综合信息库。尽管该项目的目标是实现人类知识的总和，但各语言版本之间存在严重的内容不平衡。为了量化和调查这些失衡，我们研究了文化语境对40个语言版本的影响。为此，我们开发了一种计算方法来识别与每个维基百科语言版本相关联的编辑文化背景相关的文章。我们采用了多种策略的组合，考虑了文章的地理位置、特定的关键词和类别以及文章之间的链接。我们通过人工评估对该方法的质量进行了验证，平均准确率为0.92，平均召回率为0.95。结果显示，每个维基百科语言版本中约有四分之一致力于代表相应的文化背景。虽然这些内容的相当一部分是在项目的头几年创建的，但它的创建会随着时间的推移而持续下去。对这一内容的跨语言报道的分析表明，其中大部分内容在其原始语言中是独一无二的，并揭示了文化背景之间的特殊联系；同时，它突出了百科全书可以扩展其内容的差距。这项研究中提出的方法和发现有助于促进维基百科的参与和跨文化丰富。生成的数据集可供进一步研究使用。
The online encyclopedia Wikipedia is the largest general information repository created through collaborative efforts from all over the globe. Despite the project's goal being to achieve the sum of human knowledge, there are strong content imbalances across the language editions. In order to quantify and investigate these imbalances, we study the impact of cultural context in 40 language editions. To this purpose, we developed a computational method to identify articles that can be related to the editors' cultural context associated to each Wikipedia language edition. We employed a combination of strategies taking into account geolocated articles, specific keywords and categories, as well as links between articles. We verified the method's quality with manual assessment and found an average precision of 0.92 and an average recall of 0.95. The results show that about a quarter of each Wikipedia language edition is dedicated to represent the corresponding cultural context. Although a considerable part of this content was created during the first years of the project, its creation is sustained over time. An analysis of cross-language coverage of this content shows that most of it is unique in its original language, and reveals special links between cultural contexts; at the same time, it highlights gaps where the encyclopedia could extend its content. The approach and findings presented in this study can help to foster participation and inter-cultural enrichment of Wikipedias. The datasets produced are made available for further research.