Mining causal rules hidden in spatial co-locations based on dynamic spatial databases
Mining causal rules hidden in spatial co-locations based on dynamic spatial databases
复制标题
DOI：
10.1109/cits.2016.7546420
复制
发表时间：
2016-07
期刊：
影响因子：
--
通讯作者：
Junli Lu;Lizhen Wang;Yuan Fang
中科院分区：
文献类型：
--
作者：
Junli Lu;Lizhen Wang;Yuan Fang
空间同位表示经常位于地理空间中的空间特征的子集。空间协同挖掘是近年来的一个研究热点。但关于隐藏在空间并置中的因果规则发现的研究尚未见报道。也许同一地点的特征偶然地共享了相似的环境，或者它们竞争地生活在相同的环境中，它们本身没有因果关系。因此，从大量的流行同位词中挖掘因果规则就更加有趣了。提出了一种基于动态空间数据库的流行并置挖掘因果规则的算法。由于大量的流行并置集合和同一并置规则的数量，发现的计算代价很高，因此提出了剪枝策略来在可接受的时间段内解决该问题。实验结果表明，因果规则仅占同位规则的60%左右，而且算法的计算能力更强。
Spatial co-locations represent the subsets of spatial features which are frequently located together in a geographic space. Spatial co-location mining has been a research hot in recent years. But the research on causal rule discovery hidden in spatial co-locations has not been reported. Maybe the features in a co-location accidentally share the similar environment, and maybe they are competitively living in the same environment, they themselves have no causal relationships. So mining causal rules in amount of prevalent co-locations is more interesting. This paper proposes a novel algorithm to mine causal rules from prevalent co-locations based on dynamic spatial databases. Because of large collections of prevalent co-locations and amount of rules in one co-location, the computational cost for the discovery is high, thus the pruning strategies are presented to solve the problem in an acceptable period of time. The extensive experiments evaluate the proposed algorithms with “real + synthetic” data sets and the results show that causal rules are just about 60% of co-location rules, and which are more powerful.