挖掘DBLP作者合作关系，FP-Growth算法实践（4）：挖掘每个会议的核心研究者

在只有【论文标题、发布时间、作者、会议名称】这四种信息的情况下，首先提取出所有这四种信息：

代码产生的结果如下，数据结构类似于headerTable，看结果就知道了，不再介绍：

authorDict={} #{authorName: total(frequence, startYear, endYear), {eachConf(frequence, startYear, endYear)}}

    authorDict, conferenceDict=CountAuthorAndConferenceFrequence(tranDB)
    print len(authorDict) #31886
    print authorDict["Ying Wu"]
    #[51, 2000, 2015, {'KDD': [1, 2007, 2007], 'SDM': [1, 2005, 2005], 'CVPR': [49, 2000, 2015]}]
    print conferenceDict
    #{'CVPR': 7722, 'PAKDD': 1760, 'ICML': 1111, 'KDD': 2360, 'COLT': 736, 'ICDM': 3873, 'DMKD': 47, 'SIGIR': 2772, 'ECML/PKDD': 1036, 'SDM': 1156, 'WSDM': 618, 'NIPS': 3852}

代码为：

def CountAuthorAndConferenceFrequence(tranDB):
    authorDict={} #{authorName: total(frequence, startYear, endYear), {eachConf(frequence, startYear, endYear)}}
    conferenceDict={} #{conference: count}
    for i, (conf, year, authorList) in enumerate(tranDB):
        print "trans", i, "=="*20
        if conf is np.nan or year is np.nan or authorList is np.nan:
            continue #for tranDB[2426,:]
            
        if conferenceDict.has_key(conf):
            conferenceDict[conf]+=1
        else:
            conferenceDict[conf]=1
        for author in authorList.split("|"):
            if authorDict.has_key(author):
                #total(frequence, startYear, endYear)
                authorDict[author][0]+=1
                if year<authorDict[author][1]:
                    authorDict[author][1]=year
                elif year>authorDict[author][2]:
                    authorDict[author][2]=year
                
                #eachConf(frequence, startYear, endYear)
                if authorDict[author][3].has_key(conf):
                    authorDict[author][3][conf][0]+=1
                    if year<authorDict[author][3][conf][1]:
                        authorDict[author][3][conf][1]=year
                    elif year>authorDict[author][3][conf][2]:
                        authorDict[author][3][conf][2]=year
                else:
                    authorDict[author][3][conf]=[1, year, year]
            else:
                authorDict[author]=[1, year, year, {}]
                authorDict[author][3][conf]=[1, year, year]
    return authorDict, conferenceDict

提取核心研究者，就是阈值控制，没技术含量：

'''
authorDict={} #{authorName: total(frequence, startYear, endYear), {eachConf(frequence, startYear, endYear)}}
conferenceDict={} #{conference: count}
'''
def FindCoreResearcher(authorDict, conferenceDict, coreSupport=0.05):
    wf=open("CoreResearcher.txt","w")
    
    allConferenceTotalCount=0
    for conf, count in conferenceDict.items():
        allConferenceTotalCount+=count
        coreSup=count*coreSupport
        if coreSup<2: #at least 2, or how can we say you are a core researcher?
            coreSup=2
        elif coreSup>20: #some conference find to few core researchers
            coreSup=coreSup/2
        elif coreSup>10: #some conference find to few core researchers
            coreSup=10 #but is we choose coreSup/2, will to many, so we select this threshold
        print "conf:", conf, ",total paper count:", count, "##"*20, "coreSup=", coreSup
        wf.write("conf:"+conf+",total paper count:"+str(count)+"##"*20+"coreSup="+str(coreSup)+"\n")
        for author in authorDict.keys():
            if authorDict[author][3].has_key(conf) and authorDict[author][3][conf][0]>=coreSup:
                print "%s is a core researcher with support %s" % (author, authorDict[author][3][conf][0])
                wf.write(author+" is a core researcher with support "+str(authorDict[author][3][conf][0])+"\n")
                wf.write("\t His/Her active time is==>["+str(authorDict[author][3][conf][1])+"-"+str(authorDict[author][3][conf][2])+"]\n")
    
    paperCount_authorCount_Dict={}
    coreSup=allConferenceTotalCount*coreSupport
    if coreSup>len(conferenceDict)*15: 
        coreSup=len(conferenceDict)*15/4 #at least a paper one year in (at least 1/4 of all the conferences)
    print "all conference, total paper count:", allConferenceTotalCount, "##"*20, "the overall coreSupport=", coreSup
    wf.write("all conference, total paper count:"+str(allConferenceTotalCount)+"##"*20+"the overall coreSupport="+str(coreSup)+"\n")
    for author in authorDict.keys():
        if authorDict[author][0]>=coreSup:
            print "%s is a core researcher with support %s" % (author, authorDict[author][0])
            wf.write(author+" is a core researcher with support "+str(authorDict[author][0])+"\n")
            wf.write("\t His/Her active time is==>["+str(authorDict[author][1])+"-"+str(authorDict[author][2])+"]\n")
                
        #count how many authors have published "authorDict[author][0]" papers
        #this is for the frequent pattern minSupport
        if paperCount_authorCount_Dict.has_key(authorDict[author][0]):
            paperCount_authorCount_Dict[authorDict[author][0]]+=1
        else:
            paperCount_authorCount_Dict[authorDict[author][0]]=1
            
    wf.close()
    return paperCount_authorCount_Dict