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).trim())).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         },
  36         appendCandidateWordList: function(strTag){
  37                 var s = this.env.memory.candidateWordList.isIncluded(strTag, function(a, b){ return (a.str == b.str); });
  38                 if(s){
  39                         s.wordCount++;
  40                 } else{
  41                         strTag.wordCount = 1;
  42                         this.env.memory.appendMemoryTag(strTag);
  43                 }
  44         },
  45         mergeCandidateWordList: function(strTagList){
  46                 for(var i = 0, iLen = strTagList.length; i < iLen; i++){
  47                         this.appendCandidateWordList(strTagList[i]);
  48                 }
  49         },
  50         debugShowCandidateWordList: function(){
  51                 var c = (new Array()).concat(this.env.memory.candidateWordList);
  52                 c.reverse();
  53                 this.env.debug("candidateWordList:" + c.length + "\n");
  54
  55                 for(var i = 0, iLen = c.length; i < iLen; i++){
  56                         this.env.debug((i + 1) + ":\t" + c[i].wordCount.toString() + ":\t" + c[i].wordLevel.toString() + ":\t" + c[i].str + "\n");
  57                 }
  58                 this.env.debug("candidateWordList end\n");
  59         },
  60         filterCandidateWordList00:function(cList){
  61                 //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
  62                 //cList内の候補単語に対して、フィルターをかける。
  63                 var iLen = cList.length;
  64                 if(iLen < 1){
  65                         return;
  66                 }
  67                 var baseStrTag = cList[0];
  68                 for(var i = 1; i < iLen; i++){
  69                         var c = cList[i];
  70                         if(baseStrTag.str.indexOf(c.str) != -1){
  71                                 //c.strはbaseStrTag.strに含まれている
  72                                 if(baseStrTag.wordCount == c.wordCount){
  73                                         //かつ出現回数が等しいので不要な単語
  74                                         //後で削除する。出現回数を0にマークする。
  75                                         c.wordCount = 0;
  76                                 }
  77                         }
  78                         if(c.wordCount > 0){
  79                                 //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
  80                                 var baseStrTag = c;
  81                         }
  82                 }
  83                 //削除処理
  84                 for(var i = 1; i < iLen; i++){
  85                         var c = cList[i];
  86                         if(c.wordCount == 0){
  87                                 cList.removeByIndex(i);
  88                                 i--;
  89                                 iLen--;
  90                         }
  91                 }
  92         },
  93         filterCandidateWordList01:function(cList, minLen){
  94                 //01:minLenに満たない文字数の候補を削除
  95                 //削除処理
  96                 var iLen = cList.length;
  97                 for(var i = 0; i < iLen; i++){
  98                         if(cList[i].str.length < minLen){
  99                                 cList.removeByIndex(i);
 100                                 i--;
 101                                 iLen--;
 102                         }
 103                 }
 104         },
 105         filterCandidateWordList02:function(cList, minCount){
 106                 //02:minCountに満たない出現回数の候補を削除
 107                 //削除処理
 108                 var iLen = cList.length;
 109                 for(var i = 0; i < iLen; i++){
 110                         if(cList[i].wordCount < minCount){
 111                                 cList.removeByIndex(i);
 112                                 i--;
 113                                 iLen--;
 114                         }
 115                 }
 116         },
 117         sortCandidateWordListByWordCount: function(){
 118                 this.env.memory.candidateWordList.stableSort(function(a, b){
 119                         return a.wordCount - b.wordCount;
 120                 });
 121         },
 122         sortCandidateWordListByWordLevel: function(){
 123                 this.env.memory.candidateWordList.stableSort(function(a, b){
 124                         return a.wordLevel - b.wordLevel;
 125                 });
 126         },
 127         computeWordLevel: function(strTag){
 128                 var s = strTag.str;
 129                 var iLen = s.length;
 130                 var f = 0;
 131                 strTag.wordLevel = 0;
 132                 //文字列中の文字種数を数える
 133                 for(var i = 0; i < iLen; i++){
 134                         if(s.isHiraganaAt(i)){
 135                                 f |= 0x01;
 136                         } else if(s.isKanjiAt(i)){
 137                                 f |= 0x02;
 138                         } else if(s.isKatakanaAt(i)){
 139                                 f |= 0x04;
 140                         } else if(s.isHankakuKanaAt(i)){
 141                                 f |= 0x08;
 142                         } else{
 143                                 f |= 0x10;
 144                         }
 145                 }
 146                 for(var i = 0; i < 5; i++){
 147                         if((f & 0x01) != 0){
 148                                 strTag.wordLevel++;
 149                         }
 150                         f >>>= 1;
 151                 }
 152                 strTag.wordLevel = 1 / strTag.wordLevel;
 153                 return;
 154         },
 155         computeEachWordLevel: function(){
 156                 var iLen = this.env.memory.candidateWordList.length;
 157                 for(var i = 0; i < iLen; i++){
 158                         this.computeWordLevel(this.env.memory.candidateWordList[i]);
 159                 }
 160         }
 161 }