组合查询条件下的属性社区搜索
ATTRIBUTED COMMUNITY SEARCH METHOD UNDER COMBINED QUERY CONDITION
-
摘要: 传统的属性社区搜索问题仅研究查询属性在结果社区中存在与否。为应对复杂查询场景的需求, 研究了组合查询条件下的属性社区搜索问题:给定多属性集合, 各属性至少需满足的数量以及社区大小的上限约束, 搜索所有节点的最小度数最大化的社区。提出通用算法解决框架, 并以此为基础, 提出两个优化方法, 分别是:基于属性特征的搜索空间优化, 以减小搜索空间; 基于结构特征的搜索顺序优化, 以通过搜索顺序的调整进一步提升算法效率。实验结果表明, 算法可找到符合组合查询条件的属性社区。在大规模数据集上, 经过两个优化后的算法效率比原算法提升2~3倍, 同时内存开销减少约50%。Abstract: The traditionally attributed community search problem only studies whether the query attribute exists in the resulting community. To address the needs of complex query scenarios, attributed community search problem under combined query condition is studied. We gave multiple attribute sets, the minimum number of each attribute and the upper limit of community size, and we searched for a community with the maximal number of the minimum degree of nodes. This paper proposed a general algorithm solution framework and two optimization methods: search space optimization based on attribute features to reduce the search space; search order optimization based on structural features to improve algorithm efficiency further by adjusting the search order. Experimental results show that the algorithm can find the attributed community that meets the combined query condition. After two optimizations, the optimized algorithm’s efficiency is 2~3 times higher than the original algorithm on large datasets, and memory overhead is reduced by about 50%.