Research On The Partition Techniques For Large-scale Evolution Graph

The graph is a kind of abstract data structure that is often used in computerscience. The universality of the graph makes the actual network in the real world canoften be abstracted into a graph data model. At present, it has been widely used incomputer science, linguistics, logic, physics, and chemistry, telecommunicationengineering and other fields. However, with the rapid development of the networkand computer technology, and the rapid growth of the real network scale, the scale ofthe graph grows increasingly. Meanwhile, they also aggravate the degree of thedynamic evolution of the actual network, leading to the constant evolution of graphover time. So how to dispose the large-scale evolution graph efficiently has becomethe research focus and difficulty in recent years.This paper researches deeply the existing graph partition techniques. Althoughsome existing techniques can be applied to the large-scale graph partition efficiently,the existing research of the graph partition technology is still mainly aiming at thestatic graph partition that has been unable to satisfy the large-scale graph partitionunder the dynamic evolution.According to the above problem, this paper mainly studies the problems of thepartition method of the large-scale dynamic evolution graph and the dynamicmaintenance of the large-scale evolution graph partition. Firstly, this paper proposesa large-scale evolution graph partition method based on the increment. It is able topartition the increment graph data of the large-scale evolution graph efficiently, andreduces the computational overhead and communication overhead effectively,improving the efficiency of partition. Secondly, this paper proposes a dynamicmaintenance strategy based on the closeness to realize the dynamic maintenance ofthe large-scale evolution graph partition. In order to avoid partitioning the entiregraph again, this paper only adjusts the partial graph data to realize the dynamicmaintenance of the result of the large-scale evolution graph partition. It can ensurethe effect of partition, at the same time improves the execution efficiency of the method. Finally, this paper verifies the method in time performance and validity viaa mass of experiments. The experimental results show that the method of thelarge-scale evolution graph partition that this paper proposed can dispose thepartition of the large-scale graph under dynamic evolution efficiently and has fastspeed. In addition, this paper proposes the dynamic maintenance strategy of thelarge-scale evolution graph partition can realize the dynamic maintenance of theresult of the large-scale evolution graph partition, and it can ensure the veracity ofthe partition result.