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                         }
  68                         if(this.env.memory.candidateWordList[i].wordLevel < 1){
  69                                 this.env.debug("Too small wordLevel of candidateWord [" + this.env.memory.candidateWordList[i].str + "]. Removed.\n");
  70                                 this.env.memory.removeMemoryTagByObject(this.env.memory.candidateWordList[i]);
  71                                 i--;
  72                                 iLen--;
  73                         }
  74                 }
  75                 this.env.memory.candidateWordListLastCleanedDate = new Date();
  76         },
  77         debugShowCandidateWordList: function(){
  78                 var c = this.env.memory.candidateWordList.copy();
  79                 c.reverse();
  80                 this.env.debug("candidateWordList:" + c.length + "\n #:wCount:level:str\n");
  81
  82                 for(var i = 0, iLen = c.length; i < iLen; i++){
  83                         this.env.debug((i + 1) + ":\t" + c[i].wordCount.toString() + ":\t" + c[i].wordLevel.toString() + ":\t" + c[i].str + "\n");
  84                 }
  85                 this.env.debug("candidateWordList end\n");
  86         },
  87         filterCandidateWordList00:function(cList){
  88                 //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
  89                 //cList内の候補単語に対して、フィルターをかける。
  90                 var iLen = cList.length;
  91                 if(iLen < 1){
  92                         return;
  93                 }
  94                 var baseStrTag = cList[0];
  95                 for(var i = 1; i < iLen; i++){
  96                         var c = cList[i];
  97                         if(baseStrTag.str.indexOf(c.str) != -1){
  98                                 //c.strはbaseStrTag.strに含まれている
  99                                 if(baseStrTag.wordCount == c.wordCount){
 100                                         //かつ出現回数が等しいので不要な単語
 101                                         //後で削除する。出現回数を0にマークする。
 102                                         c.wordCount = 0;
 103                                 }
 104                         }
 105                         if(c.wordCount > 0){
 106                                 //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
 107                                 var baseStrTag = c;
 108                         }
 109                 }
 110                 //削除処理
 111                 for(var i = 1; i < iLen; i++){
 112                         var c = cList[i];
 113                         if(c.wordCount == 0){
 114                                 cList.removeByIndex(i);
 115                                 i--;
 116                                 iLen--;
 117                         }
 118                 }
 119         },
 120         filterCandidateWordList01:function(cList, minLen){
 121                 //01:minLenに満たない文字数の候補を削除
 122                 var iLen = cList.length;
 123                 for(var i = 0; i < iLen; i++){
 124                         if(cList[i].str.length < minLen){
 125                                 cList.removeByIndex(i);
 126                                 i--;
 127                                 iLen--;
 128                         }
 129                 }
 130         },
 131         filterCandidateWordList02:function(cList, minCount){
 132                 //02:minCountに満たない出現回数の候補を削除
 133                 var iLen = cList.length;
 134                 for(var i = 0; i < iLen; i++){
 135                         if(cList[i].wordCount < minCount){
 136                                 cList.removeByIndex(i);
 137                                 i--;
 138                                 iLen--;
 139                         }
 140                 }
 141         },
 142         filterCandidateWordList03: function(cList){
 143                 //03:すでに単語と判明している候補を削除
 144                 var iLen = cList.length;
 145                 for(var i = 0; i < iLen; i++){
 146                         if(this.env.memory.getUUIDFromWord(cList[i].str) != this.env.UUID_Meaning_UndefinedString){
 147                                 cList.removeByIndex(i);
 148                                 i--;
 149                                 iLen--;
 150                         }
 151                 }
 152         },
 153         sortCandidateWordListByWordCount: function(){
 154                 this.env.memory.candidateWordList.stableSort(function(a, b){
 155                         return a.wordCount - b.wordCount;
 156                 });
 157         },
 158         sortCandidateWordListByWordLevel: function(){
 159                 this.env.memory.candidateWordList.stableSort(function(a, b){
 160                         return a.wordLevel - b.wordLevel;
 161                 });
 162         },
 163         sortWordListByLength: function(){
 164                 //文字数の大きい方がリストの最初に来るようにする。
 165                 this.env.memory.wordList.stableSort(function(a, b){
 166                         return b.str.length - a.str.length;
 167                 });
 168         },
 169         computeWordLevel: function(strTag){
 170                 var s = strTag.str;
 171                 var iLen = s.length;
 172                 var f = 0;
 173                 strTag.wordLevel = 0;
 174                 //文字列中の文字種数を数える
 175                 for(var i = 0; i < iLen; i++){
 176                         if(s.isHiraganaAt(i)){
 177                                 f |= 0x01;
 178                         } else if(s.isKanjiAt(i)){
 179                                 f |= 0x02;
 180                         } else if(s.isKatakanaAt(i)){
 181                                 f |= 0x04;
 182                         } else if(s.isHankakuKanaAt(i)){
 183                                 f |= 0x08;
 184                         } else{
 185                                 f |= 0x10;
 186                         }
 187                 }
 188                 for(var i = 0; i < 5; i++){
 189                         if((f & 0x01) != 0){
 190                                 strTag.wordLevel++;
 191                         }
 192                         f >>>= 1;
 193                 }
 194                 strTag.wordLevel = 1 / strTag.wordLevel;
 195                 return;
 196         },
 197         computeEachWordLevel: function(){
 198                 var iLen = this.env.memory.candidateWordList.length;
 199                 for(var i = 0; i < iLen; i++){
 200                         this.computeWordLevel(this.env.memory.candidateWordList[i]);
 201                 }
 202         },
 203         splitByWord: function(s){
 204                 if(!this.wordListCache || this.wordListCacheLastModifiedDate < this.env.memory.wordListLastModifiedDate){
 205                         //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
 206                         this.sortWordListByLength();
 207                         this.wordListCache = this.env.memory.wordList.propertiesNamed("str");
 208                         this.wordListCacheLastModifiedDate = new Date();
 209                 }
 210                 return s.splitByArraySeparatorSeparatedLong(this.wordListCache);
 211         },
 212         getUUIDListFromSeparatedString: function(separated){
 213                 var retv = new Array();
 214                 for(var i = 0, iLen = separated.length; i < iLen; i++){
 215                         retv.push(this.env.memory.getUUIDFromWord(separated[i]));
 216                 }
 217                 return retv;
 218         },
 219 }