资源描述：
《人工智能：智能导医系统中TFIDF权重改进算法研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、2382017，53（4）ComputerEngineeringandApplications计算机工程与应用智能导医系统中TF-IDF权重改进算法研究111，21徐奕枫，刘利军，黄青松，傅铁威111，21XUYifeng,LIULijun,HUANGQingsong,FUTiewei1.昆明理工大学信息工程与自动化学院，昆明6505002.昆明理工大学云南省计算机技术应用重点实验室，昆明6505001.FacultyofInformationEngineeringandAutomation,KunmingU
2、niversityofScienceandTechnology,Kunming650500,China2.YunnanKeyLaboratoryofComputerTechnologyApplications,KunmingUniversityofScienceandTechnology,Kunming650500,ChinaXUYifeng,LIULijun,HUANGQingsong,etal.ResearchonTF-IDFweightimprovementalgorithminintelligentg
3、uidancesystem.ComputerEngineeringandApplications,2017,53（4）：238-243.Abstract：Intelligentguidancesystemorientedtopatientsusingartificialintelligencetechnology,accordingthepatient’ssymptomstocalculatepossibledisease,guidethepatientstoregistrarscorrectly.Atpre
4、sent,mostoftheintelligentguidesystemthroughinputthepatient’ssymptomsorputforwardquestionsinthewebsite.Thisapproachispronetoappearproblemswhichthepatient’sinputandmedicalprofessionalsymptomsofthewordsdonotmatch,resultingthediseaseofcalculatedlowreliability.I
5、nordertosolvethisproblem,thispaperputsforwardthebarycenterbackwardalgorithmcombinationwithmedicalprofessionsthesaurustotheliteralsimilarityalgorithmtoidentifysynonyms,mappingthesymptomstothesymptomsofthepatient;Accordingtothesymptomsofthedisease,whetherimpo
6、rtantornotonlyineachofthecharacteristicsofadisease,proposethecalculationmethodofuserattentionbasedonthefrequencyofsymptom;thetraditionalTF-IDFalgorithmtakesproblemwhichsymptomsuneveninclassifydiseases,putforwardsymptomsweightimprovingalgorithmbasedondisease
7、distribution.Theexperimentresultsshowthattheproposedandimprovedalgorithmhasbettereffectonperformance.Keywords：artificialintelligencesystem;artificialintelligence;barycenterbackwardalgorithm;synonymsimilaritymatching;TF-IDFalgorithm摘要：面向患者的智能导医系统通过人工智能技术，依据患
8、者症状计算可能疾病，引导患者准确挂号。目前智能导医系统多采用患者输入描述自身症状或者提问的方式，该方式易出现患者输入与医学专业症状词不匹配的问题，导致计算出的疾病可信度较低。针对这一问题，提出重心后移和医学专业语料库相结合的方法，对同义词匹配，映射出与患者症状对应的症状词；根据症状不论重要与否在每一疾病中仅出现一次的特点，提出基于患者关注度的症状词频计算方法；针对传统TF-IDF算法在待分类疾病类中
