面向本地和外地用户情感分析推荐模型.pdfVIP

2021年河北大学学报(自然科学版)2021

第41卷第4期JournalofHebeiUniversity(NaturalScienceEdition)Vol.41No.4

DOI：10.3969/j.in.1000-1565.2021.04.012

面向本地和外地用户情感分析推荐模型

魏宁I,袁方1刘宇I

(1.河北大学数学与信息科学学院，河北保定071002；2.河北大学计算机教学部.河北保定071002)

摘要：针对地理标签和评论信息的情感倾向对于推荐系统性能的影响.本文基于地理标签和用户评论

情感分析提出有关兴趣点的推荐策略，并建立了一种基于内容的推荐模型.本系统首先对用户兴趣点信息进

行有效的补充.并实现了用户兴趣点相似度度量.对无标签评论数据进行情感分析及挖掘.获取其情感倾向

度.同时本系统结合了时间滑动窗口，更准确地把握用户评论和兴趣点的结合度.最终得到用户个性化推荐

排名.本文方法涵盖了本地用户和外地用户的个性化推荐策略.通过实验数据表明，本文模型有效提高了推

荐的准确度.

关键词：地理位置；基于内容的推荐；情感分析；循环神经网络

中图分类号:TP391文献标志码：A文章编号:1000-1565(2021)04-0419-07

Recommendationmodelofsentimentanalysis

forbothhome-townandout-of-townusers

WEINing1,YUANFang12,LIUYu1

(1.CollegeofMathematicandInformationScience,HebeiUniversity,Baoding071002,China；

2.ComputerScienceTeachingDepartment,HebeiUniversity,Baoding071002,China)

Abstract:Aimingattheinfluenceofthesentimenttendencyofgeographicalpositionandcommenton

theperformanceoftherecommendationsystem,thipaperproposearecommendationstrategyforpoints-

of-interestbasedongeographicalpositionandsentimentanalysiofuserreviews,andestablishea

content-basedrecommendationmodel.First,thesystemeffectivelysupplementtheuserpoint-of-interest

information,andrealizethesimilaritymeasurementoftheuserpoint-of-interest.Secondly,emotional

analysiandminingofunlabeledcommentdataarecarriedouttoobtainthesentimentaltendency.Atthe

sametime,thesystemcombineatimeslidingwindow,andmoreaccuratelygraspthecombinationofuser

commentandpoints-of-interest.Finally,thepersonalizedrecommendationrankingofuseriob