dist/trunk/rec10/auto_rec.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009-2011 Yukikaze
   5 import n_gram
   6 import rec10d
   7 def getDBKey(key,chtxt):
   8     return rec10d.rec10db.select_by_key_in_auto_bayes_key(key, chtxt)
   9 def changeDBEpg(chtxt,beforenum,newnum):
  10     rec10d.rec10db.change_ratio_all_reduce_in_auto_bayes_key(chtxt, beforenum, newnum)
  11 def changeDBNum(chtxt,recaddnum,alladdnum):
  12     """
  13     add num.(Not change)
  14     """
  15     rec10d.rec10db.add_num_in_auto_bayes_key(chtxt, recaddnum,alladdnum)
  16 def changeDBRatioRec(key,chtxt,beforenum,addnum):
  17     rec10d.rec10db.change_ratio_rec_in_auto_bayes_key(key, chtxt, beforenum, addnum)
  18 def changeDBRatioMultiRec(chtxt,beforenum,list):
  19     rec10d.rec10db.change_multi_ratio_rec_in_auto_bayes_key(chtxt, beforenum, list)
  20 def changeDBReduceRatioRec(chtxt,beforenum,addnum):
  21     rec10d.rec10db.change_ratio_rec_reduce_in_auto_bayes_key(chtxt, beforenum, addnum)
  22 def changeDBRatioAll(key,chtxt,beforenum,addnum):
  23     rec10d.rec10db.change_ratio_all_in_auto_bayes_key(key, chtxt, beforenum, addnum)
  24 def changeDBRatioMultiAll(chtxt,beforenum,list):
  25     rec10d.rec10db.change_multi_ratio_all_in_auto_bayes_key(chtxt,beforenum, list)
  26 def getDBNum(chtxt):
  27     ret=getDBKey("NUM", chtxt)
  28     if len(ret)<3:
  29         rec10d.rec10db.add_in_auto_bayes_key("NUM", chtxt, 1,1)
  30         ret=getDBKey("NUM", chtxt)
  31     return ret
  32 def updateRecall(chtxt,titles,descs,newnum):
  33     """
  34     update recall
  35     """
  36     titles=" "+titles+" "
  37     descs=" "+descs+" "
  38     str1=titles*2+descs+titles*2
  39     noun=n_gram.getNounQuadGram(str1)
  40     bnum=getDBNum(chtxt)[3]
  41     if bnum<1:
  42         bnum=100
  43     changeDBRatioMultiAll(chtxt, bnum,noun)
  44     changeDBEpg(chtxt, bnum, newnum)
  45     changeDBNum(chtxt, 0, newnum)
  46 def addKey(chtxt,title,desc):
  47     title=u" "+title+u" "
  48     desc=u" "+desc+u" "
  49     str1=title*2+desc+title*2##タイトルは重視したいので幾度か足す。
  50     noun=n_gram.getNounQuadGram(str1)#nounは辞書のキーに文字を、中身に出現回数を書いたもの。
  51     bnum=getDBNum(chtxt)[2]
  52     if bnum<1:
  53         bnum=1
  54     changeDBRatioMultiRec(chtxt, bnum, noun)
  55     changeDBNum(chtxt,1,0)
  56     changeDBReduceRatioRec(chtxt, bnum,1)
  57 def calcKey(chtxt,title,desc):
  58     """
  59
  60     """
  61     title=u" "+title+u" "
  62     desc=u" "+desc+u" "
  63     str1=title*2+desc+title*2##タイトルは重視したいので幾度か足す。
  64     noun=n_gram.getNounQuadGram(str1)#nounは辞書のキーに文字を、中身に出現回数を書いたもの。
  65     tnum=1
  66     nump=0
  67     for key,num in noun.items():
  68         pp=getDBKey(key,chtxt)
  69         pp2=getDBKey(key,"ALL")
  70         if len(pp)>3:
  71             tarec=0
  72             taall=0
  73             if len(pp2)>3:
  74                 tarec=pp2[2]
  75                 taall=pp2[3]
  76             trec=pp[2]
  77             tall=pp[3]
  78
  79             tnum=tnum+1
  80             p=(1000*trec+5)/(1000*tall+5)
  81             p=p*p*100
  82             p2=(1000*tarec+5)/(1000*taall+5)
  83             p2=p2*100
  84             nump=nump+p+p2
  85     return int(nump/tnum)