aiwrcgnz.js

   1 function AI_WordRecognition(env){
   2         this.env = env;
   3         this.wordListCache = null;
   4         this.wordListCacheLastModifiedDate = new Date();
   5 }
   6 AI_WordRecognition.prototype = {
   7         slideLookUpCandidateWordByHistory: function(input){
   8                 var h = this.env.input.historyList;
   9                 var cList = new Array();
  10                 for(var i = 0, iLen = input.length; i < iLen; i++){
  11                         //input character loop
  12                         var iStr = input.substr(i);
  13                         var cLen = 0;
  14                         var cStr = "";
  15                         for(var j = 0, jLen = h.length; j < jLen; j++){
  16                                 //history entry loop
  17                                 var hStrBase = h[j];
  18                                 for(var k = 0, kLen = hStrBase.length; k < kLen; k++){
  19                                         //history character loop
  20                                         var hStr = hStrBase.substr(k);
  21                                         var m = hStr.compareLeftHand(iStr);
  22                                         if(m > cLen && m != iStr.length){
  23                                                 cLen = m;
  24                                         }
  25                                 }
  26                         }
  27                         if(cLen > 0){
  28                                 cList.pushUnique(new AI_CandidateWordTag(iStr.substr(0, cLen).trim())).wordCount++;
  29                         }
  30                 }
  31                 //フィルター
  32                 this.filterCandidateWordList00(cList);
  33                 this.filterCandidateWordList01(cList, 2);
  34                 this.filterCandidateWordList03(cList);
  35                 //追加
  36                 this.mergeCandidateWordList(cList);
  37
  38         },
  39         appendCandidateWordList: function(strTag){
  40                 var s = this.env.memory.candidateWordList.isIncluded(strTag, function(a, b){ return (a.str == b.str); });
  41                 if(s){
  42                         s.wordCount++;
  43                 } else{
  44                         strTag.wordCount = 1;
  45                         this.env.memory.appendMemoryTag(strTag);
  46                 }
  47         },
  48         getCandidateWordTagByString: function(str){
  49                 return this.env.memory.candidateWordList.isIncluded(str, function(a, b){ return (a.str == b); });
  50         },
  51         mergeCandidateWordList: function(strTagList){
  52                 for(var i = 0, iLen = strTagList.length; i < iLen; i++){
  53                         this.appendCandidateWordList(strTagList[i]);
  54                 }
  55         },
  56         cleanCandidateWordList: function(){
  57                 //不要な候補単語を削除
  58                 //出現回数の少ない候補単語
  59                 //単語度が1未満の単語(暫定)
  60                 var iLen = this.env.memory.candidateWordList.length;
  61                 for(var i = 0; i < iLen; i++){
  62                         if(this.env.memory.candidateWordList[i].wordCount < 10){
  63                                 this.env.debug("Too small wordCount of candidateWord [" + this.env.memory.candidateWordList[i].str + "]. Removed.\n");
  64                                 this.env.memory.removeMemoryTagByObject(this.env.memory.candidateWordList[i]);
  65                                 i--;
  66                                 iLen--;
  67                                 continue;
  68                         }
  69                         if(this.env.memory.candidateWordList[i].wordLevel < 1){
  70                                 this.env.debug("Too small wordLevel of candidateWord [" + this.env.memory.candidateWordList[i].str + "]. Removed.\n");
  71                                 this.env.memory.removeMemoryTagByObject(this.env.memory.candidateWordList[i]);
  72                                 i--;
  73                                 iLen--;
  74                                 continue;
  75                         }
  76                 }
  77                 this.env.memory.candidateWordListLastCleanedDate = new Date();
  78         },
  79         debugShowCandidateWordList: function(){
  80                 var c = this.env.memory.candidateWordList.copy();
  81                 c.reverse();
  82                 this.env.debug("candidateWordList:" + c.length + "\n #:wCount:level:str\n");
  83
  84                 for(var i = 0, iLen = c.length; i < iLen; i++){
  85                         this.env.debug((i + 1) + ":\t" + c[i].wordCount.toString() + ":\t" + c[i].wordLevel.toString() + ":\t" + c[i].str + "\n");
  86                 }
  87                 this.env.debug("candidateWordList end\n");
  88         },
  89         filterCandidateWordList00:function(cList){
  90                 //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
  91                 //cList内の候補単語に対して、フィルターをかける。
  92                 var iLen = cList.length;
  93                 if(iLen < 1){
  94                         return;
  95                 }
  96                 var baseStrTag = cList[0];
  97                 for(var i = 1; i < iLen; i++){
  98                         var c = cList[i];
  99                         if(baseStrTag.str.indexOf(c.str) != -1){
 100                                 //c.strはbaseStrTag.strに含まれている
 101                                 if(baseStrTag.wordCount == c.wordCount){
 102                                         //かつ出現回数が等しいので不要な単語
 103                                         //後で削除する。出現回数を0にマークする。
 104                                         c.wordCount = 0;
 105                                 }
 106                         }
 107                         if(c.wordCount > 0){
 108                                 //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
 109                                 var baseStrTag = c;
 110                         }
 111                 }
 112                 //削除処理
 113                 for(var i = 1; i < iLen; i++){
 114                         var c = cList[i];
 115                         if(c.wordCount == 0){
 116                                 cList.removeByIndex(i);
 117                                 i--;
 118                                 iLen--;
 119                         }
 120                 }
 121         },
 122         filterCandidateWordList01:function(cList, minLen){
 123                 //01:minLenに満たない文字数の候補を削除
 124                 var iLen = cList.length;
 125                 for(var i = 0; i < iLen; i++){
 126                         if(cList[i].str.length < minLen){
 127                                 cList.removeByIndex(i);
 128                                 i--;
 129                                 iLen--;
 130                         }
 131                 }
 132         },
 133         filterCandidateWordList02:function(cList, minCount){
 134                 //02:minCountに満たない出現回数の候補を削除
 135                 var iLen = cList.length;
 136                 for(var i = 0; i < iLen; i++){
 137                         if(cList[i].wordCount < minCount){
 138                                 cList.removeByIndex(i);
 139                                 i--;
 140                                 iLen--;
 141                         }
 142                 }
 143         },
 144         filterCandidateWordList03: function(cList){
 145                 //03:すでに単語と判明している候補を削除
 146                 var iLen = cList.length;
 147                 for(var i = 0; i < iLen; i++){
 148                         if(this.env.memory.getUUIDFromWord(cList[i].str) != this.env.UUID_Meaning_UndefinedString){
 149                                 cList.removeByIndex(i);
 150                                 i--;
 151                                 iLen--;
 152                         }
 153                 }
 154         },
 155         sortCandidateWordListByWordCount: function(){
 156                 this.env.memory.candidateWordList.stableSort(function(a, b){
 157                         return a.wordCount - b.wordCount;
 158                 });
 159         },
 160         sortCandidateWordListByWordLevel: function(){
 161                 this.env.memory.candidateWordList.stableSort(function(a, b){
 162                         return a.wordLevel - b.wordLevel;
 163                 });
 164         },
 165         sortWordListByLength: function(){
 166                 //文字数の大きい方がリストの最初に来るようにする。
 167                 this.env.memory.wordList.stableSort(function(a, b){
 168                         return b.str.length - a.str.length;
 169                 });
 170         },
 171         computeWordLevel: function(strTag){
 172                 var s = strTag.str;
 173                 var iLen = s.length;
 174                 var f = 0;
 175                 strTag.wordLevel = 0;
 176                 //文字列中の文字種数を数える
 177                 for(var i = 0; i < iLen; i++){
 178                         if(s.isHiraganaAt(i)){
 179                                 f |= 0x01;
 180                         } else if(s.isKanjiAt(i)){
 181                                 f |= 0x02;
 182                         } else if(s.isKatakanaAt(i)){
 183                                 f |= 0x04;
 184                         } else if(s.isHankakuKanaAt(i)){
 185                                 f |= 0x08;
 186                         } else{
 187                                 f |= 0x10;
 188                         }
 189                 }
 190                 for(var i = 0; i < 5; i++){
 191                         if((f & 0x01) != 0){
 192                                 strTag.wordLevel++;
 193                         }
 194                         f >>>= 1;
 195                 }
 196                 strTag.wordLevel = 1 / strTag.wordLevel;
 197                 return;
 198         },
 199         computeEachWordLevel: function(){
 200                 var iLen = this.env.memory.candidateWordList.length;
 201                 for(var i = 0; i < iLen; i++){
 202                         this.computeWordLevel(this.env.memory.candidateWordList[i]);
 203                 }
 204         },
 205         splitByWord: function(s){
 206                 if(!this.wordListCache || this.wordListCacheLastModifiedDate < this.env.memory.wordListLastModifiedDate){
 207                         //キャッシュが存在しないか古い場合、元のリストをソートしてからキャッシュを作成
 208                         this.sortWordListByLength();
 209                         this.wordListCache = this.env.memory.wordList.propertiesNamed("str");
 210                         this.wordListCacheLastModifiedDate = new Date();
 211                 }
 212                 return s.splitByArraySeparatorSeparatedLong(this.wordListCache);
 213         },
 214         getUUIDListFromSeparatedString: function(separated){
 215                 var retv = new Array();
 216                 for(var i = 0, iLen = separated.length; i < iLen; i++){
 217                         retv.push(this.env.memory.getUUIDFromWord(separated[i]));
 218                 }
 219                 return retv;
 220         },
 221 }