components/test/data/dom_distiller/derived_features.json

   1 [
   2  {
   3   "url": "https://zh-tw.facebook.com/permalink.php",
   4   "index": 0,
   5   "features": [
   6    "opengraph",
   7    false,
   8    "forum",
   9    false,
  10    "index",
  11    false,
  12    "view",
  13    false,
  14    "asp",
  15    false,
  16    "phpbb",
  17    false,
  18    "php",
  19    true,
  20    "pathlength",
  21    14,
  22    "domain",
  23    false,
  24    "pathcomponents",
  25    1,
  26    "slugdetector",
  27    1,
  28    "pathnumbers",
  29    0,
  30    "lastSegmentLength",
  31    13,
  32    "formcount",
  33    1,
  34    "anchorcount",
  35    26,
  36    "elementcount",
  37    120,
  38    "anchorratio",
  39    0.21666666666666667,
  40    "innertextlength",
  41    492,
  42    "textcontentlength",
  43    12832,
  44    "innerhtmllength",
  45    19715,
  46    "innertextlengthratio",
  47    0.024955617550088766,
  48    "textcontentlengthratio",
  49    0.6508749682982501,
  50    "innertexttextcontentlengthratio",
  51    0.03834164588528678,
  52    "innertextwordcount",
  53    5,
  54    "textcontentwordcount",
  55    1470,
  56    "innerhtmlwordcount",
  57    2313,
  58    "innertextwordcountratio",
  59    0.00216169476869866,
  60    "textcontentwordcountratio",
  61    0.6355382619974059,
  62    "innertexttextcontentwordcountratio",
  63    0.003401360544217687
  64   ]
  65  },
  66  {
  67   "url": "https://zb.getitfree.us/myaccount/survey.html",
  68   "index": 1,
  69   "features": [
  70    "opengraph",
  71    false,
  72    "forum",
  73    false,
  74    "index",
  75    false,
  76    "view",
  77    false,
  78    "asp",
  79    false,
  80    "phpbb",
  81    false,
  82    "php",
  83    false,
  84    "pathlength",
  85    22,
  86    "domain",
  87    false,
  88    "pathcomponents",
  89    2,
  90    "slugdetector",
  91    1,
  92    "pathnumbers",
  93    0,
  94    "lastSegmentLength",
  95    11,
  96    "formcount",
  97    0,
  98    "anchorcount",
  99    5,
 100    "elementcount",
 101    92,
 102    "anchorratio",
 103    0.05434782608695652,
 104    "innertextlength",
 105    1137,
 106    "textcontentlength",
 107    25037,
 108    "innerhtmllength",
 109    33602,
 110    "innertextlengthratio",
 111    0.03383727159097673,
 112    "textcontentlengthratio",
 113    0.7451044580679721,
 114    "innertexttextcontentlengthratio",
 115    0.045412789072173186,
 116    "innertextwordcount",
 117    181,
 118    "textcontentwordcount",
 119    2811,
 120    "innerhtmlwordcount",
 121    3926,
 122    "innertextwordcountratio",
 123    0.04610290371879776,
 124    "textcontentwordcountratio",
 125    0.7159959246051961,
 126    "innertexttextcontentwordcountratio",
 127    0.06438989683386695
 128   ]
 129  },
 130  {
 131   "url": "https://yande.re/pool",
 132   "index": 2,
 133   "features": [
 134    "opengraph",
 135    false,
 136    "forum",
 137    false,
 138    "index",
 139    false,
 140    "view",
 141    false,
 142    "asp",
 143    false,
 144    "phpbb",
 145    false,
 146    "php",
 147    false,
 148    "pathlength",
 149    5,
 150    "domain",
 151    false,
 152    "pathcomponents",
 153    1,
 154    "slugdetector",
 155    0,
 156    "pathnumbers",
 157    0,
 158    "lastSegmentLength",
 159    4,
 160    "formcount",
 161    10,
 162    "anchorcount",
 163    112,
 164    "elementcount",
 165    492,
 166    "anchorratio",
 167    0.22764227642276422,
 168    "innertextlength",
 169    1676,
 170    "textcontentlength",
 171    41089,
 172    "innerhtmllength",
 173    56900,
 174    "innertextlengthratio",
 175    0.02945518453427065,
 176    "textcontentlengthratio",
 177    0.7221265377855888,
 178    "innertexttextcontentlengthratio",
 179    0.04078950570712356,
 180    "innertextwordcount",
 181    303,
 182    "textcontentwordcount",
 183    3874,
 184    "innerhtmlwordcount",
 185    6502,
 186    "innertextwordcountratio",
 187    0.04660104583205168,
 188    "textcontentwordcountratio",
 189    0.5958166717932943,
 190    "innertexttextcontentwordcountratio",
 191    0.07821373257614868
 192   ]
 193  }
 194 ]