Privacy preserving mining from outsourced graph database with authenticated subgraph query services

Abstract

Graph data structure has been used widely in several fields, including newlinecheminformatics, bioinformatics, communication networks, social networks, newlineweb analysis and to describe complex objects. In recent decades, because of the newlineexplosive growth of graph databases, the contribution of subgraph mining in the newlinefield of Graph mining has caught the attention of many researchers owing to its newlinewide range of applications. In large scale dynamic graph databases, mining newlinesubgraphs patterns is challenging because graph operations, such as subgraph newlineisomorphism testing are often more complex in time, memory and cost than the newlinecorresponding operations on sets, sequences, or trees. Graph indexing and newlinemining algorithms had been proposed so far are vastly different in terms of newlinecomputation time and memory requirement. These algorithms compromised newlinescalability and execution time while attempting to exploit large scale graph newlinedatabases. In current distributed environment, very few research works have newlineconcentrated on secure subgraph mining and faced lot of issues in reducing newlineauthentication time. A useful technique with scalability, security and QoS is still newlinerequired for subgraph mining. newlineThe proposed work is focussed to solve the existing issues in graph newlineindexing, subgraph searching, authentication of query service and scalability of newlinegraph data. A system model for subgraph mining with authentication of query newlineservices in graph database is proposed in three different modules for three newlinedifferent applications. In the first module, an efficient authentication friendly newlinestatic graph index called GSimIndex is constructed to support authenticated newlinesearch for similarity subgraph queries based on Distance-based indexing. This newlinework is tested for AIDS Antiviral screening dataset and compared with existing newlinegraph indexing techniques that supports similarity search like DISR index, newlineGMTree and ML-Index. newline