Font Size: a A A

Research Of Software Important Patterns Based On Complex Network

Posted on:2019-09-09Degree:MasterType:Thesis
Country:ChinaCandidate:X M TianFull Text:PDF
GTID:2370330566989091Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the prosperity and development of the software industry,the function of software becomes more and more abundant,which also brings great convenience to people's life and software security problems.On the basis of the complex network theory,which is of great significance both in analyzing software key nodes and analyzing the frequent traversal patterns of software.This paper treat the function of software system as a granularity unit,based on the structural features of the software network,study the influence propagation and traversal patterns characteristics.mine the important nodes and the representative important patterns in the software system.The main work is as follows.Firstly,there is uncertainty in an execution trace of the software in mapping the software system into complex network topology.this paper integrate N software execution trace,construct the weighted software network.The edge weight c in the network represents that the invoking dependency relationship appears c times in the multiple execution trace.Secondly,previous studies tended to ignore the invoking dependency relationships in measuring the importance of software node,lead to the lack of effective analysis method.For this purpose,this paper import the PageRank into software engineering field,propose an algorithm FunctionRank to measure the importance of nodes,integrate the invoking times and outdegree of the nodes into the importance measure.In the transmission of importance,according to the invoking times of the nodes,transferring in proportion,which is more consistent with the objective reality.According to the outdegree of the nodes,the probability of transmission between software nodes with higher outdegree is higher.Thirdly,the general sequential pattern mining algorithm lacks the consideration of the continuity of the software traversal patterns.For this purpose,this paper constructs frequent traversal candidate pattern tree FWCP-Tree,puts forward the frequent traversal patterns mining algorithm OSFPMiner,select the maximum node weight in all traversals of a pattern as its upper-bound,use the pruning strategy based on upper-bound in the process of constructing,mine software important patterns.Finally,the mining algorithms proposed in this paper are implemented on the Windows platform,using java language.the potential characteristics of the software system are mined and the effectiveness for the proposed methods are verified.
Keywords/Search Tags:weighted software network, important node, frequent traversal pattern, upper-bound, pruning
PDF Full Text Request
Related items