A RESTful Web Service For Non-overlapping Community Quality Assessment

Most networks in the real world have an implicit community structure.Analyzing community structures is helpful for revealing networks' hidden meaningful structures and functionality.When applying community detection to mine these community structures,it gives rise to the requirement to evaluate the quality of the detected community structure,which is called Community Quality Assessment(CQA for short).However,large scale communities detected from nowadays rapidly growing social networks present great challenges to the computation efficiency of CQA.Existing sequential algorithms have high computation complexity.Though parallel algorithms using message passing interface(MPI)have recently been introduced into the field,its computation efficiency needs further improvement.Meanwhile,the complexity of the MPI implementation handicaps data scientists from adopting it.To improve the efficiency of CQA computation and ease its usage,we propose a novel paralleled methods and embed it in a framework as Web service.We first combine the advantages of existing algorithms and proposes a faster MPI-based NonOverlapping CQA metrics computation algorithm.Then,we develop a grouping method using greedy strategy to balance among MPI cluster nodes.Thirdly,an empirical comparison study has been carried out to demonstrate the efficiency of our proposed algorithm in terms of execution time,memory consumption and speedup using various real massive communities generated from artificial networks.Finally,we propose a RESTful framework to wrap the non-overlapping CQA metrics computation as a Web service,i.e.,non-overlapping CommuMetrics,which makes it easy to use.As a proof of concept,we provide an implementation of CommuMetrics Web service prototype,which provides CQA metrics computation as Web service provision.