| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 1 | <!DOCTYPE html> |
| 2 | <html> |
| 3 | <head> |
| 4 | <title>DeReKo-Word-Vector-Distances: <%= $word %></title> |
| Marc Kupietz | 80bd7b9 | 2017-07-04 16:25:54 +0200 | [diff] [blame] | 5 | <link rel="stylesheet" href="//code.jquery.com/ui/1.12.1/themes/base/jquery-ui.css"> |
| 6 | <script src="http://code.jquery.com/jquery-latest.min.js"></script> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 7 | <script src = "https://cdn.datatables.net/1.10.16/js/jquery.dataTables.min.js"></script> |
| 8 | <script src = "https://cdn.datatables.net/fixedcolumns/3.2.3/js/dataTables.fixedColumns.min.js"></script> |
| 9 | <link rel="stylesheet" href="https://cdn.datatables.net/1.10.16/css/jquery.dataTables.min.css"> |
| Marc Kupietz | 80bd7b9 | 2017-07-04 16:25:54 +0200 | [diff] [blame] | 10 | <script |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 11 | src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js" |
| 12 | integrity="sha256-VazP97ZCwtekAsvgPBSUwPFKdrwD3unUfSGVYrahUqU=" |
| 13 | crossorigin="anonymous"></script> |
| Marc Kupietz | 80bd7b9 | 2017-07-04 16:25:54 +0200 | [diff] [blame] | 14 | <script> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 15 | $('#firstable').hide(); |
| 16 | $(document).ready(function() { |
| 17 | $("#xxxtabs").tabs( { |
| 18 | "show": function(event, ui) { |
| 19 | var oTable = $('div.dataTables_scrollBody>table.display', ui.panel).dataTable(); |
| 20 | if ( oTable.length > 0 ) { |
| 21 | oTable.fnAdjustColumnSizing(); |
| 22 | } |
| 23 | } |
| 24 | } ); |
| Marc Kupietz | 694610d | 2017-11-25 18:30:03 +0100 | [diff] [blame] | 25 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 26 | $(".selector").tabs({ active: 1 }); |
| Marc Kupietz | 0af83e3 | 2017-11-27 09:31:37 +0100 | [diff] [blame] | 27 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 28 | $('#firsttable').DataTable({ |
| 29 | "sScrollY": "780px", |
| 30 | "bScrollCollapse": true, |
| 31 | "bPaginate": false, |
| 32 | "bJQueryUI": true, |
| 33 | "dom": '<"top">rt<"bottom"flp><"clear">', |
| 34 | "aoColumnDefs": [ |
| 35 | { "sWidth": "10%", "aTargets": [ -1 ] } |
| 36 | ] |
| 37 | } ); |
| Marc Kupietz | 0af83e3 | 2017-11-27 09:31:37 +0100 | [diff] [blame] | 38 | |
| 39 | $('#secondtable').DataTable({ |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 40 | "sScrollY": "800px", |
| 41 | "bScrollCollapse": true, |
| 42 | "bPaginate": false, |
| 43 | "bJQueryUI": true, |
| 44 | "dom": '<"top">rt<"bottom"flp><"clear">', |
| 45 | "aoColumnDefs": [ |
| 46 | { "sWidth": "10%", "aTargets": [ -1 ] } |
| 47 | ] |
| 48 | } ); |
| 49 | }); |
| Marc Kupietz | 0af83e3 | 2017-11-27 09:31:37 +0100 | [diff] [blame] | 50 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 51 | $( function() { |
| 52 | $( "#iterations" ).spinner({ |
| 53 | spin: function( event, ui ) { |
| 54 | if ( ui.value < 1000 ) { |
| 55 | $( this ).spinner( "value", 1000 ); |
| 56 | return false; |
| 57 | } else if ( ui.value > 10000 ) { |
| 58 | $( this ).spinner( "value", 10000 ); |
| 59 | return false; |
| 60 | } |
| 61 | } |
| 62 | }); |
| 63 | } ); |
| Marc Kupietz | 3305b0a | 2017-11-27 10:46:20 +0100 | [diff] [blame] | 64 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 65 | $( function() { |
| 66 | $( "#neighbours" ).spinner({ |
| 67 | spin: function( event, ui ) { |
| 68 | if ( ui.value < 0 ) { |
| 69 | $( this ).spinner( "value", 0 ); |
| 70 | return false; |
| 71 | } else if ( ui.value > 200 ) { |
| 72 | $( this ).spinner( "value", 200 ); |
| 73 | return false; |
| 74 | } |
| 75 | } |
| 76 | }); |
| 77 | } ); |
| 78 | |
| 79 | $( function() { |
| 80 | $( "#cutoff" ).spinner({ |
| 81 | spin: function( event, ui ) { |
| 82 | if ( ui.value < 100000 ) { |
| 83 | $( this ).spinner( "value", 100000 ); |
| 84 | return false; |
| 85 | } else if ( ui.value > 2000000 ) { |
| 86 | $( this ).spinner( "value", 2000000 ); |
| 87 | return false; |
| 88 | } |
| 89 | } |
| 90 | }); |
| 91 | } ); |
| 92 | |
| 93 | $( function() { |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 94 | $( "#tabs" ).tabs(); // .addClass('ui-helper-clearfix tabs-left-vertical'); |
| 95 | } ); |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 96 | |
| 97 | $( function() { |
| 98 | $( ".controlgroup-vertical" ).controlgroup({ |
| 99 | "direction": "vertical" |
| 100 | }); |
| 101 | } ); |
| 102 | |
| 103 | $(function() { |
| 104 | $( document ).tooltip({ |
| 105 | content: function() { |
| 106 | return $(this).attr('title'); |
| 107 | }} |
| 108 | ) |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 109 | }) |
| Marc Kupietz | 694610d | 2017-11-25 18:30:03 +0100 | [diff] [blame] | 110 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 111 | </script> |
| 112 | <script src="//d3js.org/d3.v3.min.js" charset="utf-8"></script> |
| Marc Kupietz | 554aff5 | 2017-11-09 14:42:09 +0100 | [diff] [blame] | 113 | <script src="/derekovecs/js/tsne.js"></script> |
| 114 | <script src="/derekovecs/js/som.js"></script> |
| 115 | <script src="/derekovecs/js/labeler.js"></script> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 116 | <style> |
| 117 | body, input { |
| 118 | font-family: Arial, sans-serif; |
| 119 | font-size: 11pt; |
| 120 | } |
| Marc Kupietz | 30ca434 | 2017-11-22 21:21:20 +0100 | [diff] [blame] | 121 | |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 122 | .tabs-left-vertical .ui-tabs-nav { |
| 123 | position: absolute; |
| 124 | width: 21em; |
| 125 | transform: translate(-100%,0%) rotate(-90deg); |
| 126 | transform-origin: 100% 0%; |
| 127 | } |
| 128 | . |
| 129 | .tabs-left-vertical .ui-tabs-nav li { |
| 130 | float: right; |
| 131 | } |
| 132 | |
| 133 | .tabs-left-vertical .ui-tabs-panel { |
| 134 | padding-left: 3.5em; |
| 135 | } |
| 136 | |
| 137 | .tabs-left-vertical .ui-tabs-panel { |
| 138 | height: 20em; |
| 139 | } |
| 140 | |
| Marc Kupietz | 30ca434 | 2017-11-22 21:21:20 +0100 | [diff] [blame] | 141 | .mono { |
| 142 | font-family: "DejaVu Sans Mono", Inconsolata, SourceCodePro, Courier; |
| 143 | } |
| 144 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 145 | .ui-tooltip-content { |
| 146 | font-size: 9pt; |
| 147 | color: #222222; |
| 148 | } |
| 149 | |
| 150 | svg > .ui-tooltip-content { |
| 151 | font-size: 8pt; |
| 152 | color: #222222; |
| 153 | } |
| 154 | |
| 155 | a.merged { |
| 156 | color: green; |
| 157 | fill: green; |
| 158 | } |
| 159 | |
| 160 | #first a { |
| 161 | text-decoration: none; |
| 162 | } |
| 163 | |
| 164 | a.marked, #first a.marked { |
| 165 | text-decoration: underline; |
| 166 | } |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 167 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 168 | a.target { |
| 169 | color: red; |
| 170 | fill: red; |
| 171 | } |
| Marc Kupietz | 694610d | 2017-11-25 18:30:03 +0100 | [diff] [blame] | 172 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 173 | table.display { |
| 174 | width: 40% important!; |
| 175 | margin: 1; /* <- works for me this way ****/ |
| 176 | } |
| 177 | table.dataTable thead th, table.dataTable thead td, table.dataTable tbody td { |
| 178 | padding: 2px 2px; |
| 179 | // border-bottom: 1px solid #111; |
| 180 | } |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 181 | #collocators { |
| 182 | margin-bottom: 15px; |
| 183 | } |
| 184 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 185 | #topwrapper { |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 186 | width: 100%; |
| 187 | // border: 1px solid red; |
| 188 | overflow: hidden; /* will contain if #first is longer than #second */ |
| 189 | } |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 190 | |
| 191 | #wrapper { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 192 | // border: 1px solid red; |
| 193 | overflow: hidden; /* will contain if #first is longer than #second */ |
| 194 | } |
| 195 | |
| 196 | #options { |
| 197 | float: right; |
| 198 | margin: 20px; |
| 199 | max-width: 280px; |
| 200 | overflow: hidden; /* if you don't want #second to wrap below #first */ |
| 201 | } |
| 202 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 203 | #first { |
| 204 | margin-right: 20px; |
| 205 | float: left; |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 206 | overflow: hidden; /* if you don't want #second to wrap below #first */ |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 207 | // border: 1px solid green; |
| 208 | } |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 209 | #tabs { |
| 210 | margin-right: 20px; |
| 211 | overflow: hidden; /* if you don't want #second to wrap below #first */ |
| 212 | } |
| 213 | |
| 214 | #embed { |
| 215 | max-width: 802px; |
| 216 | border: 1px solid #333; |
| 217 | } |
| 218 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 219 | #second { |
| Marc Kupietz | b0173f1 | 2017-11-29 10:00:23 +0100 | [diff] [blame^] | 220 | min-width: 800px; |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 221 | // border: 1px solid #333; |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 222 | overflow: hidden; /* if you don't want #second to wrap below #first */ |
| 223 | } |
| 224 | #som2 svg { |
| 225 | border: 1px solid #333; |
| 226 | } |
| 227 | |
| 228 | #cost { |
| 229 | font-size: 8pt; |
| 230 | color: #222222; |
| 231 | margin-top: 4px; |
| 232 | margin-bottom: 12px; |
| 233 | } |
| 234 | |
| 235 | #sominfo1, #sominfo { |
| 236 | font-size: 8pt; |
| 237 | color: #222222; |
| 238 | margin-top: 0px; |
| 239 | } |
| 240 | |
| 241 | #somcolor1, #somcolor2, #somcolor3 { |
| 242 | display: inline-block; |
| 243 | height: 10px; |
| 244 | width: 10px; |
| 245 | } |
| 246 | |
| 247 | #third { |
| 248 | border: 1px solid #333; |
| 249 | } |
| 250 | |
| 251 | </style> |
| 252 | <script> |
| 253 | |
| 254 | var opt = {epsilon: <%= $epsilon %>, perplexity: <%= $perplexity %>}, |
| 255 | mapWidth = 800, // width map |
| 256 | mapHeight = 800, |
| 257 | jitterRadius = 7; |
| 258 | |
| 259 | var T = new tsnejs.tSNE(opt); // create a tSNE instance |
| 260 | |
| 261 | var Y; |
| 262 | |
| 263 | var data; |
| 264 | var labeler; |
| 265 | |
| 266 | function applyJitter() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 267 | svg.selectAll('.tsnet') |
| 268 | .data(labels) |
| 269 | .transition() |
| 270 | .duration(50) |
| 271 | .attr("transform", function(d, i) { |
| 272 | T.Y[i][0] = (d.x - mapWidth/2 - tx)/ss/20; |
| 273 | T.Y[i][1] = (d.y - mapHeight/2 - ty)/ss/20; |
| 274 | return "translate(" + |
| 275 | (d.x) + "," + |
| 276 | (d.y) + ")"; |
| 277 | }); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 278 | } |
| 279 | |
| 280 | function updateEmbedding() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 281 | var Y = T.getSolution(); |
| 282 | svg.selectAll('.tsnet') |
| 283 | .data(data.words) |
| 284 | .attr("transform", function(d, i) { |
| 285 | return "translate(" + |
| 286 | ((Y[i][0]*20*ss + tx) + mapWidth/2) + "," + |
| 287 | ((Y[i][1]*20*ss + ty) + mapHeight/2) + ")"; }); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 288 | } |
| 289 | |
| 290 | var svg; |
| 291 | var labels = []; |
| 292 | var anchor_array = []; |
| 293 | var text; |
| 294 | |
| 295 | function drawEmbedding() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 296 | $("#embed").empty(); |
| 297 | var div = d3.select("#embed"); |
| 298 | |
| 299 | // get min and max in each column of Y |
| 300 | var Y = T.Y; |
| 301 | |
| 302 | svg = div.append("svg") // svg is global |
| 303 | .attr("width", mapWidth) |
| 304 | .attr("height", mapHeight); |
| 305 | |
| 306 | var g = svg.selectAll(".b") |
| 307 | .data(data.words) |
| 308 | .enter().append("g") |
| 309 | .attr("class", "tsnet"); |
| 310 | |
| 311 | g.append("a") |
| 312 | .attr("xlink:href", function(word) { |
| 313 | return (data.urlprefix+word);}) |
| 314 | .attr("class", function(d, i) { |
| 315 | var res=""; |
| 316 | if(data.marked[i]) { |
| 317 | res="marked "; |
| 318 | } |
| 319 | if(data.target.indexOf(" "+d+" ") >= 0) { |
| 320 | return res+"target"; |
| 321 | } else if(data.ranks[i] < data.mergedEnd) { |
| 322 | return res+"merged"; |
| 323 | } else { |
| 324 | return res; |
| 325 | } |
| 326 | }) |
| 327 | .attr("title", function(d, i) { |
| 328 | if(data.mergedEnd > 0) { |
| 329 | if(data.ranks[i] >= data.mergedEnd) { |
| 330 | return "rank: "+i +" "+"freq. rank: "+(data.ranks[i]).toString().replace(/\B(?=(\d{3})+(?!\d))/g, ","); |
| 331 | } else { |
| 332 | return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",") + " (merged vocab)"; |
| 333 | } |
| 334 | } else { |
| 335 | return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ","); |
| 336 | } |
| 337 | }) |
| 338 | .append("text") |
| 339 | .attr("text-anchor", "top") |
| 340 | .attr("font-size", 12) |
| 341 | .text(function(d) { return d; }); |
| 342 | |
| 343 | var zoomListener = d3.behavior.zoom() |
| 344 | .scaleExtent([0.1, 10]) |
| 345 | .center([0,0]) |
| 346 | .on("zoom", zoomHandler); |
| 347 | zoomListener(svg); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 348 | } |
| 349 | |
| 350 | var tx=0, ty=0; |
| 351 | var ss=1; |
| 352 | var iter_id=-1; |
| 353 | |
| 354 | function zoomHandler() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 355 | tx = d3.event.translate[0]; |
| 356 | ty = d3.event.translate[1]; |
| 357 | ss = d3.event.scale; |
| 358 | updateEmbedding(); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 359 | } |
| 360 | |
| 361 | var stepnum = 0; |
| 362 | |
| 363 | function stopStep() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 364 | clearInterval(iter_id); |
| 365 | text = svg.selectAll("text"); |
| 366 | |
| 367 | // jitter function needs different data and co-ordinate representation |
| 368 | labels = d3.range(data.words.length).map(function(i) { |
| 369 | var x = (T.Y[i][0]*20*ss + tx) + mapWidth/2; |
| 370 | var y = (T.Y[i][1]*20*ss + ty) + mapHeight/2; |
| 371 | anchor_array.push({x: x, y: y, r: jitterRadius}); |
| 372 | return { |
| 373 | x: x, |
| 374 | y: y, |
| 375 | name: data.words[i] |
| 376 | }; |
| 377 | }); |
| 378 | |
| 379 | // get the actual label bounding boxes for the jitter function |
| 380 | var index = 0; |
| 381 | text.each(function() { |
| 382 | labels[index].width = this.getBBox().width; |
| 383 | labels[index].height = this.getBBox().height; |
| 384 | index += 1; |
| 385 | }); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 386 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 387 | |
| 388 | // setTimeout(updateEmbedding, 1); |
| 389 | // setTimeout( |
| 390 | labeler = d3.labeler() |
| 391 | .label(labels) |
| 392 | .anchor(anchor_array) |
| 393 | .width(mapWidth) |
| 394 | .height(mapHeight) |
| 395 | .update(applyJitter); |
| 396 | // .start(1000); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 397 | |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 398 | iter_id = setInterval(jitterStep, 1); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 399 | } |
| 400 | |
| 401 | var jitter_i=0; |
| 402 | |
| 403 | function jitterStep() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 404 | if(jitter_i++ > 100) { |
| 405 | clearInterval(iter_id); |
| 406 | } else { |
| 407 | labeler.start2(10); |
| 408 | applyJitter(); |
| 409 | } |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 410 | } |
| 411 | |
| 412 | var last_cost=1000; |
| 413 | |
| 414 | function step() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 415 | var i = T.iter; |
| 416 | |
| 417 | if(i > <%= $no_iterations %>) { |
| 418 | stopStep(); |
| 419 | } else { |
| 420 | var cost = Math.round(T.step() * 100000) / 100000; // do a few steps |
| 421 | $("#cost").html("tsne iteration " + i + ", cost: " + cost.toFixed(5)); |
| 422 | if(i % 250 == 0 && cost >= last_cost) { |
| 423 | stopStep(); |
| 424 | } else { |
| 425 | last_cost = cost; |
| 426 | updateEmbedding(); |
| 427 | } |
| 428 | } |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 429 | } |
| 430 | |
| 431 | function showMap(j) { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 432 | data=j; |
| 433 | T.iter=0; |
| 434 | T.initDataRaw(data.vecs); // init embedding |
| 435 | drawEmbedding(); // draw initial embedding |
| 436 | |
| 437 | if(iter_id >= 0) { |
| 438 | clearInterval(iter_id); |
| 439 | } |
| 440 | //T.debugGrad(); |
| 441 | iter_id = setInterval(step, 1); |
| 442 | if(<%= $show_som %>) { |
| 443 | makeSOM(j, <%= $no_iterations %>); |
| 444 | } |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 445 | } |
| Marc Kupietz | 39179ab | 2017-07-04 16:28:06 +0200 | [diff] [blame] | 446 | var queryword; |
| 447 | |
| 448 | function onload() { |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 449 | queryword = document.getElementById('word'); |
| Marc Kupietz | 39179ab | 2017-07-04 16:28:06 +0200 | [diff] [blame] | 450 | } |
| 451 | |
| 452 | function queryKorAP() { |
| 453 | window.open('http://korap.ids-mannheim.de/kalamar/?q='+queryword.value, 'KorAP'); |
| 454 | } |
| Marc Kupietz | 4dc270c | 2017-11-24 10:17:12 +0100 | [diff] [blame] | 455 | |
| 456 | function queryKorAPCII(query) { |
| 457 | window.open('http://korap.ids-mannheim.de/kalamar/?ql=cosmas2&q='+query, 'KorAP'); |
| 458 | } |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 459 | </script> |
| 460 | </head> |
| Marc Kupietz | 39179ab | 2017-07-04 16:28:06 +0200 | [diff] [blame] | 461 | <body onload="onload()"> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 462 | <div id="topwrapper"> |
| 463 | <div id="options" class="widget"> |
| 464 | <form> |
| 465 | <input id="word" type="text" name="word" size="20" placeholder="Word(s) to be searched" value="<%= $word %>" |
| 466 | title="When looking for multiple words use spaces as separators to search around the average vector and | as separator to get the neighbours for each word."/> |
| 467 | <input type="submit" value="SEARCH"> |
| 468 | <fieldset> |
| 469 | <legend>Settings</legend> |
| 470 | <div class="controlgroup-vertical"> |
| 471 | <label for="cutoff">cut-off</label> |
| 472 | <input id="cutoff" type="text" name="cutoff" size="10" value="<%= $cutoff %>" title="Only consider the most frequent x word forms."> |
| 473 | <label for="dedupe">dedupe</label> |
| 474 | <input id="dedupe" type="checkbox" name="dedupe" value="1" <%= ($dedupe ? "checked" : "") %> title="radically filter out any near-duplicates"> |
| 475 | % if($mergedEnd > 0) { |
| 476 | <label for="sbf">backw.</label> |
| 477 | <input id="sbf" type="checkbox" name="sbf" value="1" <%= ($searchBaseVocabFirst ? "checked" : "") %> title="If checkecked base vocabulary will be searched first. Otherwise merged vocabulray will be searched first."> |
| 478 | % } |
| 479 | <label for="neighbours">max. neighbours:</label> |
| 480 | <input id="neighbours" size="4" name="n" value="<%= $no_nbs %>"> |
| 481 | <label for="iterations">max. iterations</label> |
| 482 | <input id="iterations" name="N" size="4" value="<%= $no_iterations %>"> |
| 483 | <label for="dosom">SOM</label> |
| 484 | <input id="dosom" type="checkbox" name="som" value="1" <%= ($show_som ? "checked" : "") %>> |
| 485 | % if($collocators) { |
| 486 | <label for="sortby">window/sort</label> |
| 487 | <select id="sortby" name="sort"> |
| 488 | <option value="0" <%= ($sort!=1 && $sort!=2? "selected":"") %>>auto focus</option> |
| 489 | <option value="1" <%= ($sort==1? "selected":"") %>>any single position</option> |
| 490 | <option value="2" <%= ($sort==2? "selected":"") %>>whole window</option> |
| 491 | </select> |
| 492 | % } |
| 493 | <span> </span><input type="button" value="→ KorAP" onclick="queryKorAP();" title="query word with KorAP"/> |
| 494 | <br> |
| 495 | </div> |
| 496 | </fieldset> |
| 497 | </form> |
| Marc Kupietz | 0af83e3 | 2017-11-27 09:31:37 +0100 | [diff] [blame] | 498 | </div> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 499 | <div id="tabs"> |
| 500 | <ul> |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 501 | <li><a href="#tabs-1">Semantics (TSNE-map)</a></li> |
| 502 | <li><a href="#tabs-2">Semantics (SOM)</a></li> |
| 503 | <li><a href="#tabs-3">Syntagmatic (collocators)</a></li> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 504 | </ul> |
| 505 | <div id="tabs-1"> |
| 506 | % if($lists && (@$lists) > 0 && (@$lists)[0]) { |
| 507 | <div id="wrapper"> |
| 508 | <div id="first" style="width:220px"> |
| 509 | <table class="display compact nowrap" id="firsttable"> |
| 510 | <thead> |
| 511 | <tr> |
| 512 | <th align="right">#</th><th align="right">cos</th><th align="left">paradigmatic</th> |
| 513 | </tr> |
| 514 | </thead> |
| 515 | <tbody> |
| 516 | % my $j=0; my @words; my @vecs; my @ranks; my @marked; |
| 517 | % for my $list (@$lists) { |
| 518 | % my $i=0; while($list) { |
| 519 | % my $item = (@$list)[$i]; |
| 520 | % my $c = ($collocators? (@$collocators)[$i] : 0); |
| 521 | % last if(!$c && !$item); |
| 522 | <tr> |
| 523 | <td align="right"> |
| 524 | <%= ++$i %>. |
| 525 | </td> |
| 526 | % if($item) { |
| 527 | % if(!grep{$_ eq $item->{word}} @words) { |
| 528 | % push @vecs, $item->{vector}; |
| 529 | % push @words, $item->{word}; |
| 530 | % push @ranks, $item->{rank}; |
| 531 | % push @marked, ($marked->{$item->{word}}? 1 : 0); |
| 532 | % } |
| 533 | <td align="right"> |
| 534 | <%= sprintf("%.3f", $item->{dist}) %> |
| 535 | </td> |
| 536 | <td> |
| 537 | % my $class = ($marked->{$item->{word}}? "marked " : ""); |
| 538 | % my $r = $item->{rank}; |
| 539 | % if($r < $mergedEnd) { |
| 540 | % $class .= "merged"; |
| 541 | % $r .= " (merged vocab)"; |
| 542 | % } elsif($mergedEnd!=0 && $r > $mergedEnd) { |
| 543 | % $r -= $mergedEnd; |
| 544 | % } |
| 545 | <a class="<%= $class =%>" |
| 546 | title="freq. rank: <%= $r =%>" |
| 547 | href="<%= url_with->query([word => $item->{word}]) =%>"> |
| 548 | <%= $item->{word} =%> |
| 549 | </a> |
| 550 | </td> |
| 551 | % } else { |
| 552 | <td colspan="2"/> |
| 553 | % } |
| 554 | </tr> |
| 555 | % last if($i >= 100); |
| 556 | % } |
| 557 | % } |
| 558 | </tbody> |
| 559 | </table> |
| 560 | </div> |
| 561 | <script> |
| 562 | % use Mojo::ByteStream 'b'; |
| 563 | % my $urlprefix = url_with->query([word=>'']); |
| 564 | $(window).load(function() { |
| 565 | showMap(<%= b(Mojo::JSON::to_json({target => " $word ", mergedEnd=> $mergedEnd, words => \@words, vecs => \@vecs, ranks => \@ranks, marked => \@marked, urlprefix => $urlprefix})); %>); |
| 566 | }); |
| 567 | </script> |
| 568 | % } else { |
| 569 | <div id="wrapper"> |
| 570 | <p> |
| 571 | ERROR: "<%= $word %>" not found in vocabluary. |
| 572 | </p> |
| 573 | </div> |
| 574 | % } |
| 575 | <div id="second"> |
| 576 | <div id="embed"> |
| 577 | </div> |
| 578 | <div id="cost"> |
| 579 | </div> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 580 | </div> |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 581 | </div> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 582 | </div> |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 583 | <div id="tabs-2"> |
| 584 | <div id="som2" style="width: 800; height: 800px"> |
| 585 | </div> |
| 586 | <div id="sominfo1"><span id="somcolor1"> </span> <span id="somword1"> </span> <span id="somcolor2"> </span> <span id="somword2"> </span> <span id="somcolor3"> </span></div> |
| 587 | <div id="sominfo">SOM iteration <span id="iterations">0</span></div> |
| 588 | </div> |
| 589 | <div id="tabs-3"> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 590 | <div id="second" style="width:500px"> |
| 591 | <table class="display compact nowrap" id="secondtable"> |
| 592 | <thead> |
| 593 | <tr> |
| 594 | % if($collocators) { |
| 595 | <th>#</th> |
| 596 | <th align="right" title="The window around the target word that is considered for summation.">w'</th> |
| 597 | <th align="right" title="Raw (max.) activation of the collocator in the output layers.">a</th> |
| 598 | <th title="Σp(c<sub><small>@</small></sub>) – Sum of the probability approximations that the combination of the target word and the collocator at the relative position @ come from the training corpus. Single approximations can be distorted because of sub-sampling frequent words and the sum cannot itself be interpreted as probability." align="right">Σp</th> |
| 599 | <th align="right">Σp/|w|</th> |
| 600 | <th title="c" align="left">collocator</th> |
| 601 | % } |
| 602 | </tr> |
| 603 | </thead> |
| 604 | <tbody> |
| 605 | % for(my $i=0; $i < 100; $i++) { |
| 606 | % my $c = ($collocators? (@$collocators)[$i] : 0); |
| 607 | <tr> |
| 608 | <td align="right"> |
| 609 | <%= $i %> |
| 610 | </td> |
| 611 | % if($c) { |
| 612 | <td align="right"> |
| 613 | <span class="mono"><%= bitvec2window( $c->{pos} ) %></span> |
| 614 | </td> |
| 615 | <td align="right"> |
| 616 | <%= sprintf("%.3f", $c->{dist}) %> |
| 617 | </td> |
| 618 | <td align="right"> |
| 619 | <%= sprintf("%.3e", $c->{norm}) %> |
| 620 | </td> |
| 621 | <td align="right"> |
| 622 | <%= sprintf("%.3e", $c->{sum}) %> |
| 623 | </td> |
| 624 | <td align="left"> |
| 625 | <a onclick="<%= sprintf("queryKorAPCII('%s /w5 %s')", $c->{word}, $word) =%>" |
| 626 | title="freq. rank: <%= $c->{rank} =%>"> |
| 627 | <%= $c->{word} %> |
| 628 | </a> |
| 629 | </td> |
| 630 | % } else { |
| 631 | <td colspan="5"/> |
| 632 | % } |
| 633 | </tr> |
| 634 | % } |
| 635 | </tbody> |
| 636 | </table> |
| 637 | </div> <!-- - tab2 --> |
| Marc Kupietz | 4fcda0c | 2017-11-29 09:00:31 +0100 | [diff] [blame] | 638 | </div> <!-- tabs --> |
| 639 | </div> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 640 | </div> <!-- topwrapper --> |
| 641 | <div style="clear: both;"></div> |
| Marc Kupietz | 0af83e3 | 2017-11-27 09:31:37 +0100 | [diff] [blame] | 642 | </div> |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 643 | % if($training_args) { |
| 644 | <p> |
| 645 | Word vector model trained with <a href="https://code.google.com/p/word2vec/">word2vec</a> using the following parameters: <pre><%= $training_args %></pre> |
| 646 | </p> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 647 | % } |
| Marc Kupietz | 4abcd68 | 2017-11-28 20:51:08 +0100 | [diff] [blame] | 648 | </body> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 649 | </html> |