def trainSimofUsrDict(err_sim_file):
err_corpus=open(err_sim_file,'r')
line1=err_corpus.readline()
while line1:
line2=err_corpus.readline()
err_corpus.readline() # skip the empty line
line1=line1.split()[1:]
line2=line2.split()[1:]
w2vword2sim.getSimofWordListPairMod(line1,line2,modify=True,usrDict=True)
line1=err_corpus.readline()
w2vword2sim.word_sim_dic.save_dict()
def getSimNewsCorpus(news_file,sim_news_file):
corpus=list(open(news_file,'r'))
corpus_len=len(corpus)
fout=open(sim_news_file,'w')
sim_arr=np.zeros((corpus_len,corpus_len))
for i in xrange(len(corpus)):
wl1=corpus[i].split()[1:]
for j in xrange(i+1,len(corpus)):
wl2=corpus[j].split()[1:]
sim=w2vword2sim.getSimofWordListPairMod(wl1, wl2, False,False)
sim_arr[i][j]=sim_arr[j][i]=sim
sim_list=[(j,sim_arr[i][j]) for j in xrange(corpus_len) ]
sim_list=sorted(sim_list,key=lambda x:x[1],reverse=True)
print >> fout,corpus[i].strip()
for num,sim in sim_list[:10]:
print >> fout,' %.4f %s'%(sim,corpus[num].strip())
def getSimErrofRepNews(rep_file,rep_id_file,rep_sim_file,err_sim_file):
rep_corpus={}
fout=open(rep_sim_file,'w')
ferr=open(err_sim_file,'w')
for line in open(rep_file):
line=line.split()
rep_corpus[line[0]]=line[1:]
for line in open(rep_id_file):
line=line.split()
for mtid_i in line:
print >> fout,'%s %s'%(mtid_i,' '.join(rep_corpus[mtid_i]))
for mtid_j in line:
if mtid_i==mtid_j: #skip itself
continue
sim=w2vword2sim.getSimofWordListPairMod(rep_corpus[mtid_i], rep_corpus[mtid_j], modify=False,usrDict=True)
print >> fout,' %.4f %s %s'%(sim,mtid_j,' '.join(rep_corpus[mtid_j]))
if sim<0.78:
print >> ferr,'%s %s'%(mtid_i,' '.join(rep_corpus[mtid_i]))
print >> ferr,'%s %s'%(mtid_j,' '.join(rep_corpus[mtid_j]))
print >> ferr,sim
def getSimNewsRep(rep_file,news_file,sim_news_file,useusrDict=False):
corpus=list(open(news_file,'r'))
corpus_len=len(corpus)
rep=list(open(rep_file,'r'))
rep_len=len(rep)
fout=open(sim_news_file,'w')
for i in xrange(rep_len):
sim_list=[]
line=rep[i].split()
mtid,wl1=line[0],line[1:] # note this wl1 will be deleted in the getSimofWordListPairMod(wl1,wl2)
for j in xrange(corpus_len):
wl2=corpus[j].split()[1:]
sim=w2vword2sim.getSimofWordListPairMod(wl1, wl2, modify=False,usrDict=useusrDict)
sim_list.append((j,sim))
sim_list=sorted(sim_list,key=lambda x:x[1],reverse=True)
print >> fout,rep[i].strip()
for num,sim in sim_list[:11]:
if corpus[num].split()[0]==mtid: #skip the news itself
continue
print >> fout,' %.4f %s'%(sim,corpus[num].strip())
