论文标题
缺失路径：分析知识图中的不完整性
The Missing Path: Analysing Incompleteness in Knowledge Graphs
论文作者
论文摘要
知识图（kg）允许尽管有差异，但仍可以合并并连接异质数据；它们是不完整的。但是，KG数据生产商需要尽可能确保最佳的完整性水平。困难是他们没有能力区分不完整实体可以并且应该固定的案件。我们提出了一个新的可视化工具：缺失的路径，以支持它们确定可以修复的实体的相干子集。它依靠地图，根据其不完整的配置文件对实体进行分组。该地图与直方图和堆叠图表协调，以支持交互式探索和分析。可以将子集的摘要与完整集合之一进行比较，以揭示其独特的特征。我们对9个Wikidata贡献者进行了迭代设计过程和评估。参与者获得见解，并找到各种策略来确定要固定的连贯子集。
Knowledge Graphs (KG) allow to merge and connect heterogeneous data despite their differences; they are incomplete by design. Yet, KG data producers need to ensure the best level of completeness, as far as possible. The difficulty is that they have no means to distinguish cases where incomplete entities could and should be fixed. We present a new visualisation tool: The Missing Path, to support them in identifying coherent subsets of entities that can be repaired. It relies on a map, grouping entities according to their incomplete profile. The map is coordinated with histograms and stacked charts to support interactive exploration and analysis; the summary of a subset can be compared with the one of the full collection to reveal its distinctive features. We conduct an iterative design process and evaluation with 9 Wikidata contributors. Participants gain insights and find various strategies to identify coherent subsets to be fixed.
