吉林大学学报(理学版)2019,Vol.57Issue(3):583-590,8.DOI:10.13413/j.cnki.jdxblxb.2017078
基于节点相似性和链接次数组合时间序列的链接预测
LinkPredictionBasedonTimeSeriesCombined from NodeSimilaritiesandLinkNumber
摘要
Abstract
Aimingattheproblemthattheexistingmethodsmadeuseoftherelativefragmentationof networkinformation,andit was difficultto describetherelationship betweenlink numberand similarityscore,weproposedalinkprediction methodindynamicnetworks,whichusedtimeseries combinedfrom nodesimilaritiesscoresandlink mumbertopredict.Firstly,itpredictedsimilarity scoresofalltimesnapshotsbycommunityevolution.Secondly,nodessimilaritieswerecombinedwith reallinknumberbybinarytimeseriesmodelandtheprobabilityoflinkbetweenallnodepairsinnext periodwaspredicted.Finally,thetestwascarriedoutonadatasetforwardedby Weibo-Net-Tweet microblog.Theexperimentalresultsshowthatthe methodimprovesthepredictionaccuracybyat least5%,provestheintrinsicrelationshipbetweencommunityevolutionandlinkprediction,and verifiestheeffectivenessofbinarytimeseriesmodel.关键词
链接预测/社区演化/时间序列/节点相似性Key words
linkprediction/communityevolution/timeseries/nodesimilarity分类
信息技术与安全科学引用本文复制引用
魏晓辉,许国威,王兴旺,徐海啸..基于节点相似性和链接次数组合时间序列的链接预测[J].吉林大学学报(理学版),2019,57(3):583-590,8.基金项目
国家自然科学基金(批准号:61602205 ()
51627805 ()
61170004)、国家重点研发计划专项基金(批准号:2016YFB0201503 ()
2016YFB0701101)、教育部高等学校博士学科点专项科研基金(批准号:20130061110052)、吉林省科技攻关计划重大科技招标专项基金(批准号:20160203008GX)和吉林省科技攻关计划重点科技攻关项目(批准号:20140204013GX). (批准号:20130061110052)