aiwrcgnz.js

   1 function AI_WordRecognition(env){
   2         this.env = env;
   3 }
   4 AI_WordRecognition.prototype = {
   5         slideLookUpCandidateWordByHistory: function(input){
   6                 var h = this.env.input.historyList;
   7                 var cList = new Array();
   8                 for(var i = 0, iLen = input.length; i < iLen; i++){
   9                         //input character loop
  10                         var iStr = input.substr(i);
  11                         var cLen = 0;
  12                         var cStr = "";
  13                         for(var j = 0, jLen = h.length; j < jLen; j++){
  14                                 //history entry loop
  15                                 var hStrBase = h[j];
  16                                 for(var k = 0, kLen = hStrBase.length; k < kLen; k++){
  17                                         //history character loop
  18                                         var hStr = hStrBase.substr(k);
  19                                         var m = hStr.compareLeftHand(iStr);
  20                                         if(m > cLen && m != iStr.length){
  21                                                 cLen = m;
  22                                         }
  23                                 }
  24                         }
  25                         if(cLen > 0){
  26                                 cList.pushUnique(new AI_CandidateWordTag(iStr.substr(0, cLen))).wordCount++;
  27                         }
  28                 }
  29                 //フィルター
  30                 this.filterCandidateWordList00(cList);
  31                 this.filterCandidateWordList01(cList, 2);
  32                 //追加
  33                 this.mergeCandidateWordList(cList);
  34         },
  35         appendCandidateWordList: function(strTag){
  36                 var s = this.env.memory.candidateWordList.isIncluded(strTag, function(a, b){ return (a.str == b.str); });
  37                 if(s){
  38                         s.wordCount++;
  39                 } else{
  40                         strTag.wordCount = 1;
  41                         this.env.memory.appendMemoryTag(strTag);
  42                 }
  43         },
  44         mergeCandidateWordList: function(strTagList){
  45                 for(var i = 0, iLen = strTagList.length; i < iLen; i++){
  46                         this.appendCandidateWordList(strTagList[i]);
  47                 }
  48         },
  49         debugShowCandidateWordList: function(){
  50                 this.env.debug("candidateWordList:\n");
  51                 var c = this.env.memory.candidateWordList;
  52                 for(var i = 0, iLen = c.length; i < iLen; i++){
  53                         this.env.debug(c[i].wordCount.toString() + " :" + c[i].wordLevel.toString() + " :" + c[i].str + "\n");
  54                 }
  55                 this.env.debug("candidateWordList end\n");
  56         },
  57         filterCandidateWordList00:function(cList){
  58                 //00:長い単語に含まれており、かつ出現頻度が長い単語と等しい単語を削除
  59                 //cList内の候補単語に対して、フィルターをかける。
  60                 var iLen = cList.length;
  61                 if(iLen < 1){
  62                         return;
  63                 }
  64                 var baseStrTag = cList[0];
  65                 for(var i = 1; i < iLen; i++){
  66                         var c = cList[i];
  67                         if(baseStrTag.str.indexOf(c.str) != -1){
  68                                 //c.strはbaseStrTag.strに含まれている
  69                                 if(baseStrTag.wordCount == c.wordCount){
  70                                         //かつ出現回数が等しいので不要な単語
  71                                         //後で削除する。出現回数を0にマークする。
  72                                         c.wordCount = 0;
  73                                 }
  74                         }
  75                         if(c.wordCount > 0){
  76                                 //単語は削除されなかった、つまり異なる単語なので、baseStrTagを更新
  77                                 var baseStrTag = c;
  78                         }
  79                 }
  80                 //削除処理
  81                 for(var i = 1; i < iLen; i++){
  82                         var c = cList[i];
  83                         if(c.wordCount == 0){
  84                                 cList.removeByIndex(i);
  85                                 i--;
  86                                 iLen--;
  87                         }
  88                 }
  89         },
  90         filterCandidateWordList01:function(cList, minLen){
  91                 //01:minLenに満たない文字数の候補を削除
  92                 //削除処理
  93                 var iLen = cList.length;
  94                 for(var i = 0; i < iLen; i++){
  95                         if(cList[i].str.length < minLen){
  96                                 cList.removeByIndex(i);
  97                                 i--;
  98                                 iLen--;
  99                         }
 100                 }
 101         },
 102         filterCandidateWordList02:function(cList, minCount){
 103                 //02:minCountに満たない出現回数の候補を削除
 104                 //削除処理
 105                 var iLen = cList.length;
 106                 for(var i = 0; i < iLen; i++){
 107                         if(cList[i].wordCount < minCount){
 108                                 cList.removeByIndex(i);
 109                                 i--;
 110                                 iLen--;
 111                         }
 112                 }
 113         },
 114         sortCandidateWordListByWordCount: function(){
 115                 this.env.memory.candidateWordList.stableSort(function(a, b){
 116                         return a.wordCount - b.wordCount;
 117                 });
 118         },
 119         sortCandidateWordListByWordLevel: function(){
 120                 this.env.memory.candidateWordList.stableSort(function(a, b){
 121                         return a.wordLevel - b.wordLevel;
 122                 });
 123         },
 124         computeWordLevel: function(strTag){
 125                 var s = strTag.str;
 126                 var iLen = s.length;
 127                 var f = 0;
 128                 strTag.wordLevel = 0;
 129                 //文字列中の文字種数を数える
 130                 for(var i = 0; i < iLen; i++){
 131                         if(s.isHiraganaAt(i)){
 132                                 f |= 0x01;
 133                         } else if(s.isKanjiAt(i)){
 134                                 f |= 0x02;
 135                         } else if(s.isKatakanaAt(i)){
 136                                 f |= 0x04;
 137                         } else if(s.isHankakuKanaAt(i)){
 138                                 f |= 0x08;
 139                         } else{
 140                                 f |= 0x10;
 141                         }
 142                 }
 143                 for(var i = 0; i < 5; i++){
 144                         if((f & 0x01) != 0){
 145                                 strTag.wordLevel++;
 146                         }
 147                         f >>>= 1;
 148                 }
 149                 strTag.wordLevel = 1 / strTag.wordLevel;
 150                 return;
 151         },
 152         computeEachWordLevel: function(){
 153                 var iLen = this.env.memory.candidateWordList.length;
 154                 for(var i = 0; i < iLen; i++){
 155                         this.computeWordLevel(this.env.memory.candidateWordList[i]);
 156                 }
 157         }
 158 }