activeknn.py

def knnclassifier(posknn,negknn,token,k):
	
  tokenl=len(token)
	#print negknn
	
  checkp = False
  checkn = False
  prob=[0.0,0.0]

  posprob=[]
  negprob=[]

  for rev in posknn:
    posl = len(rev)
    freq=0.0
    for t in token:
        if(t in rev): 
				#print t," ",r
          checkp = True
          #print t," ",r
          freq +=1
    if (posl>0 and tokenl>0):
      freq=freq/(posl*tokenl*1.0)
      posprob.append(freq)


  for rev in negknn:
    negl = len(rev)
    freq=0.0
    for t in token:
      if(t in rev):
				#print t," ",r
          checkn = True
					#print t," ",r
          freq +=1
		#freq=freq/(negl*tokenl)
#    negprob.append(freq)
    if (negl>0 and tokenl>0):
      freq=freq/(negl*tokenl*1.0)
      negprob.append(freq)

	#print checkp," ",checkn
  posprob=sorted([w for w in posprob],reverse=True)
  negprob=sorted([w for w in negprob],reverse=True)
	#print posprob
	#print negprob
	
	#print checkn," ",checkp

  posindex=0
  negindex=0
  #print checkp,checkn
  if(checkp==1 or checkn==1):
    i=0
    while(i<k):
      if(posprob[posindex]>negprob[negindex]):
        prob[1]=prob[1]+posprob[posindex]*1.0
        posindex +=1
      else:
        prob[0]=prob[0]+negprob[i]*1.0
        negindex+=1
      i+=1
    prob.append(1)
  else:
    prob.append(0)

  return prob

def insert(l,number):
  i = len(l)-1
  j=0
  if(number>l[0]):
    while(number>l[j] and j<=i-1):
      l[j] = l[j+1]
      j+=1    
    l[j] = number
  
def knnclassifier2(posknn,negknn,tokens,k):
  sim1 = []
  sim2 = []
  if(len(tokens)==0):
    return -1
  for i in range(0,k):
    sim1.append(0.0)
    sim2.append(0.0)
  for rev in posknn:
    if(len(rev)==0):
      continue
    freq=0
    for w in tokens:
      if w in rev:
        freq+=1
    similarity = (freq*1.0)/(len(rev)*len(tokens))
    if(freq>0):
      insert(sim1,similarity)
  for rev in negknn:
    if(len(rev)==0):
      continue
    freq=0
    for w in tokens:
      if w in rev:
        freq+=1
    similarity = (freq*1.0)/(len(rev)*len(tokens))
    if(freq>0):
      insert(sim2,similarity)
  prob1 = 0.0
  prob2 = 0.0
  for i in range(0,k):
    prob1+=sim1[i]
    prob2+=sim2[i]
  if(prob1>prob2):
    return 1
  elif(prob2>prob1):
    return 0
  else:
    return -1