Abstract: Aiming at the conflict between single attribute value and spatiotemporal data regularity, the paper constructs a compact peripheral set which is closely related to the discriminated location, and adopts the Kriging method to fuse spatiotemporal data attributes. Thus, the time dimension and space dimension information are deduced to the discriminated location, so that the difference between transformed value and attribute value can be used to measure whether spatiotemporal data violates the regularity of entity objects. For avoiding the one-sidedness of the traditional method of vector space analysis, the paper puts forward a conflict detection method of spatiotemporal data based on the Pettitt test with rank sequences. It provides a solution for the conflict detection of spatiotemporal data in a certain time window. The results show that this algorithm can be applied to describe the degree of difference between the feature value of discriminated location and the estimated value of peripheral locations around the discriminated time. The average detection accuracy is up to 80%, which improves the applicability of the conflict detection algorithm for such spatiotemporal data.

Key words: spatiotemporal data, conflict detection, Kriging method, Pettitt test

摘要： 针对单个属性值与时空数据规律性之间的冲突问题，构建与判别点位密切相关的紧致周边点集，采用Kriging方法实现时空数据属性值汇集，将时间维度和空间维度信息推演到判别点位，以转化值与属性值间的差异来衡量时空数据是否违背实体对象的规律性，并通过构造秩序列的Pettitt检验给出时空数据冲突的判别方法，避免了传统依靠向量空间分析的片面性，为一定时间窗口下时空数据冲突判别提供了一个解决方案。实验表明：该方法能够有效刻画判别点位属性值与周边位置、前后时刻估计值之间差异程度，平均判别正确率可达80%，进一步提高了此类时空数据冲突判别算法的适用性。

关键词: 时空数据, 冲突判别, Kriging方法, Pettitt检验