MultiRI：标号多重图的快速子图匹配
MultiRI: Fast Subgraph Matching in Labeled Multigraphs
摘要：
子图匹配（SM）问题包括找到一个给定的小图（称为查询）到一个大图（称为目标）的所有嵌入。SM问题已被广泛研究的简单图，即两个节点之间只有一条边，节点有单一的标签，但很少有方法已经设计了标记的多图，即可能有多个标签的节点，其中对节点可能有多个标记的边缘之间的图。在这里，我们提出了MultiRI，一个新的算法的子多重图匹配（SMM）的问题，即在标记的多重图的子图匹配。MultiRI通过计算查询节点上的兼容域和对称破缺条件来过滤可能解决方案的搜索空间，从而改进了最先进的技术。从经验上讲，我们表明，MultiRI优于最先进的方法的SMM问题在合成和真实的图形，与乘法加速之间的5和10的大型图形，通过使用有限的内存量。
英文摘要：
The Subgraph Matching (SM) problem consists of finding all the embeddings of a given small graph, called the query, into a large graph, called the target. The SM problem has been widely studied for simple graphs, i.e. graphs where there is exactly one edge between two nodes and nodes have single labels, but few approaches have been devised for labeled multigraphs, i.e. graphs having possibly multiple labels on nodes in which pair of nodes may have multiple labeled edges between them. Here we present MultiRI, a novel algorithm for the Sub-Multigraph Matching (SMM) problem, i.e. subgraph matching in labeled multigraphs. MultiRI improves on the state-of-the-art by computing compatibility domains and symmetry breaking conditions on query nodes to filter the search space of possible solutions. Empirically, we show that MultiRI outperforms the state-of-the-art method for the SMM problem in both synthetic and real graphs, with a multiplicative speedup between five and ten for large graphs, by using a limited amount of memory.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
