基于图结构的相似度检索是当前图数据管理与信息检索的重要研究课题之一，它在推荐系统、金融安全、物联网、生物结构分析、知识图谱、文献计量学等各个领域已有广阔的应用前景与社会经济效益。本项目针对图数据具有海量规模、高度动态、不确定性的特点，重点研究如何在分布式集群上对海量图结构相似度进行快速、精确、并行的高性能检索，形成一套能效优化的时空复杂度低、规模可扩展性强、并发效率高的分布式处理框架与海量图相似度的高性能计算理论。拟解决的关键科学问题包括四个方面：1)分布式集群中海量规模的动态图自适应划分；2)集群节点上高效的动态索引与不同查询类型的并行检索模型；3)分布式海量图相似度检索的动态压缩技术与同步策略；4)不确定性海量动态图相似度的分布式检索与隐私保护。最后，通过研发分布式图相似度搜索引擎系统，来验证本项目中提出的理论与技术方法。本项目将为我国自主建立大数据的科学体系提供夯实的理论依据。

Similarity search based on graph topologies is one of the most important research topics in graph data management and information retrieval. It brings significant social and economic benefits, due to its wide spectrum of real emerging applications in, e.g., recommendation systems, financial security, Internet of Things, bioinformatics, knowledge graph discovery, and bibliometrics. To meet the challenges of large-scale, highly dynamical, and uncertain graph data, this project aims to develop a novel scheme for efficient graph-based similarity search over distributed systems in a fast, accurate, and parallel fashion, with the focus on taming its time and space complexities, increasing the scalability of search algorithms, and achieving the parallelism of similarity computations. There are four key scientific research problems to be addressed in this project, which encompasses: 1) adaptive partitioning of dynamical graphs over large-scale distributed systems; 2) efficient dynamic indexing on each cluster, and innovative similarity retrieval modeling to support various types of queries; 3) development of dynamical compression technologies and synchronization strategies for efficient graph-based similarity retrieval over scalable distributed systems; 4) efficient similarity search on uncertain dynamical graphs over distributed systems, with privacy-preserving techniques. Finally, a system prototype on distributed graph similarity search engine will be built to validate the effectiveness and superiority of the proposed theories, techniques and approaches. The success of this project will lay a solid theoretical foundation for the establishment of big-data information systems in our country.