blob: da669180befff871e6a2fb99f0498056b0ad33d5 [file] [log] [blame]
Marc Kupietz83305222016-04-28 09:57:22 +02001<!DOCTYPE html>
2<html>
3 <head>
4 <title>DeReKo-Word-Vector-Distances: <%= $word %></title>
Marc Kupietz80bd7b92017-07-04 16:25:54 +02005 <link rel="stylesheet" href="//code.jquery.com/ui/1.12.1/themes/base/jquery-ui.css">
Marc Kupietz6dbadd12017-11-29 16:43:33 +01006 <link href="https://fonts.googleapis.com/css?family=Lato|Roboto+Condensed" rel="stylesheet">
Marc Kupietz80bd7b92017-07-04 16:25:54 +02007 <script src="http://code.jquery.com/jquery-latest.min.js"></script>
Marc Kupietz4abcd682017-11-28 20:51:08 +01008 <script src = "https://cdn.datatables.net/1.10.16/js/jquery.dataTables.min.js"></script>
9 <script src = "https://cdn.datatables.net/fixedcolumns/3.2.3/js/dataTables.fixedColumns.min.js"></script>
Marc Kupietz22796142017-12-01 13:19:15 +010010 <script src = "https://cdn.datatables.net/plug-ins/1.10.16/sorting/scientific.js"></script>
Marc Kupietz4abcd682017-11-28 20:51:08 +010011 <link rel="stylesheet" href="https://cdn.datatables.net/1.10.16/css/jquery.dataTables.min.css">
Marc Kupietz80bd7b92017-07-04 16:25:54 +020012 <script
Marc Kupietz6dbadd12017-11-29 16:43:33 +010013 src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js"
Marc Kupietzb6c615d2017-12-02 10:38:20 +010014 integrity="sha256-VazP97ZCwtekAsvgPBSUwPFKdrwD3unUfSGVYrahUqU="
15 crossorigin="anonymous"></script>
Marc Kupietz80bd7b92017-07-04 16:25:54 +020016 <script>
Marc Kupietza6e08f02017-12-01 22:06:21 +010017 var urlParams = new URLSearchParams(window.location.search);
18
Marc Kupietz4abcd682017-11-28 20:51:08 +010019 $('#firstable').hide();
20 $(document).ready(function() {
Marc Kupietz694610d2017-11-25 18:30:03 +010021
Marc Kupietzb6c615d2017-12-02 10:38:20 +010022 $("input").bind("keydown", function(event) {
23 // track enter key
24 var keycode = (event.keyCode ? event.keyCode : (event.which ? event.which : event.charCode));
25 if (keycode == 13) { // keycode for enter key
26 // force the 'Enter Key' to implicitly click the Update button
27 document.getElementById('SEARCH').click();
28 return false;
29 } else {
30 return true;
31 }});
Marc Kupietzdab9f222017-11-29 14:22:59 +010032
Marc Kupietzb2b7aa62017-12-02 14:41:46 +010033 $( "#tabs" ).on( "tabsactivate", function( event, ui ) {
34 if (localStorage) localStorage['tab'] = ui.newTab.index();
35 });
Marc Kupietz0af83e32017-11-27 09:31:37 +010036
Marc Kupietz896c9092017-12-02 14:40:43 +010037 $(function(){
38 $("#SEARCH").click(function() {
39 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
40 });
41 });
Marc Kupietz2f6b74a2017-12-01 13:20:21 +010042
Marc Kupietzb6c615d2017-12-02 10:38:20 +010043 function changeCharColor(txt, heat) {
44 var newText = "";
45 for (var i=0, l=txt.length; i<l; i++) {
Marc Kupietz728b8ed2017-12-02 11:13:49 +010046 newText += (i == 5 ? txt.charAt(i) : '<span style="background-color:'+getHeatColor(heat[i]/maxHeat)+'">'+txt.charAt(i)+'</span>');
Marc Kupietzb6c615d2017-12-02 10:38:20 +010047 }
48 return newText;
49 }
50
51 function getHeatColor(value) {
52 var hue=((1-value)*120).toString(10);
Marc Kupietz728b8ed2017-12-02 11:13:49 +010053 return ["hsl(",hue,",90%,70%)"].join("");
Marc Kupietzb6c615d2017-12-02 10:38:20 +010054 }
55
56 function bitvec2window(n, heat) {
57 var str = n.toString(2).padStart(10, "0")
58 .replace(/^([0-9]{5})/, '$1x')
59 .replace(/0/g, '·')
60 .replace(/1/g, '+');
61 return changeCharColor(str, heat);
62 }
63
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +010064 % use Mojo::ByteStream 'b';
65 % my $urlprefix = url_with->query([word=>'']);
66 var paraResults = <%= b(Mojo::JSON::to_json($lists)) %>;
67 var urlprefix = new URLSearchParams(window.location.search);
68 if (paraResults.length > 0) {
69 var nvecs = [],
70 nwords = [],
71 nranks = [],
72 nmarked = [];
73 for(var i = 0; i < paraResults.length; i++) {
74 nwords = nwords.concat(paraResults[i].map(function(a){return a.word;}));
75 nvecs = nvecs.concat(paraResults[i].map(function(a){return a.vector;}));
76 nranks = nranks.concat(paraResults[i].map(function(a){return a.rank;}));
77 nmarked = nmarked.concat(paraResults[i].map(function(a){return a.marked;}));
78 }
79 showMap({target: " "+urlParams.get('word')+" ", mergedEnd: 0, words: nwords, vecs: nvecs, ranks: nranks, marked: nmarked, urlprefix: "?"+urlprefix} );
80 var t = $('#firsttable').DataTable({
81 data: paraResults[0],
82 "sScrollY": "780px",
83 "bScrollCollapse": true,
84 "bPaginate": false,
85 "bJQueryUI": true,
86 "dom": '<"top">rt<"bottom"flp><"clear">',
87 "columns": [
88 { "data": "rank", type: "allnumeric" },
89 { "data": "dist", render: function ( data, type, row ) {return data.toFixed(3) }},
90 { "data": "word", render: function ( data, type, row ) {urlprefix.set("word", data); return '<a href="?' + urlprefix + '">' + data + '</a>' }}
91 ],
92 "columnDefs": [
93 { className: "dt-right", "targets": [0,1] },
94 { "searchable": false,
95 "orderable": false,
96 "targets": 0
97 },
98 { "orderSequence": [ "desc" ], "targets": [ 1 ] },
99 { "orderSequence": [ "asc", "desc" ], "targets": [ 2 ] },
100 ],
101 "order": [[ 1, 'desc' ]],
102 } );
103
104 t.on( 'order.dt search.dt', function () {
105 t.column(0, {order:'applied'}).nodes().each( function (cell, i) {
106 cell.innerHTML = i+1;
107 } );
108 } ).draw();
109
110 $( "#first" ).clone().prependTo( "#tabs-2" );
111
112 }
113
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100114 var collocatorData = <%= b(Mojo::JSON::to_json($collocators)) %>;
115 var maxHeat; // = Math.max.apply(Math,collocatorData.map(function(o){return o.cprob;}))
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100116
117 if (collocatorData != null) {
Marc Kupietz896c9092017-12-02 14:40:43 +0100118 maxHeat = Math.max.apply(Math,collocatorData.map(function(o){return Math.max.apply(Math,o.heat);}))
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100119 var t = $('#secondtable').DataTable({
120 data: collocatorData,
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100121 "sScrollY": "780px",
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100122 "bScrollCollapse": true,
123 "bPaginate": false,
124 "bJQueryUI": true,
125 "dom": '<"top">rt<"bottom"flp><"clear">',
126 "columns": [
127 { "data": "rank", type: "allnumeric" },
128 { "data": "pos", width: "7%", sClass: "dt-center mono compact", render: function ( data, type, row ) {return bitvec2window(data, row.heat) }},
129 { "data": "max", render: function ( data, type, row ) {return data.toFixed(3) }},
130 { "data": "conorm", render: function ( data, type, row ) {return data.toFixed(3) }},
131 { "data": "prob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
132 { "data": "cprob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
133 { "data": "overall", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
134 { "data": "word", sClass: "collocator" }
135 ],
136 "columnDefs": [
137 { className: "dt-right", "targets": [0,2,3,4,5,6] },
138 { className: "dt-center", "targets": [ 1] },
139 { "searchable": false,
140 "orderable": false,
141 "targets": 0
142 },
143 { "type": "scientific", targets: [2,3,4,5,6] },
144 { "orderSequence": [ "desc" ], "targets": [ 2, 3, 4, 5, 6 ] },
145 { "orderSequence": [ "asc", "desc" ], "targets": [ 1, 7 ] },
146 ],
147 "order": [[ 4, 'desc' ]],
Marc Kupietz4abcd682017-11-28 20:51:08 +0100148 } );
Marc Kupietz0af83e32017-11-27 09:31:37 +0100149
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100150 t.on( 'order.dt search.dt', function () {
151 t.column(0, {order:'applied'}).nodes().each( function (cell, i) {
152 cell.innerHTML = i+1;
153 } );
154 } ).draw();
Marc Kupietzddaba632017-12-02 17:30:56 +0100155
156 if (localStorage) { // let's not crash if some user has IE7
157 var index = parseInt(localStorage['tab']||'0');
158 $("#tabs").tabs({ active: index });
159 }
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100160 }
161 $("#tabs").css("visibility", "visible"); // now we can show the tabs
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100162
Marc Kupietz4abcd682017-11-28 20:51:08 +0100163 });
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100164
Marc Kupietzdab9f222017-11-29 14:22:59 +0100165 $(function(){
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100166 $("#dropdownoptions").dialog({
167 title: "Options",
168 autoOpen: false,
169 modal: false,
170 draggable: false,
171 height: "auto",
172 width: "auto",
173 resizable: false,
174 buttons: {
175 "Cancel": function() {
176 $( this ).dialog( "close" );
177 },
178 "Apply": function() {
Marc Kupietzdab9f222017-11-29 14:22:59 +0100179 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100180 }
181 }
182 });
Marc Kupietzdab9f222017-11-29 14:22:59 +0100183 });
184
185 $(function(){
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100186 $("td.collocator").click(function(){
187 queryKorAPCII(this.textContent + " /w5 " + urlParams.get('word'));
188 });
189 });
190
191 $(function(){
192 $("#showoptions").click(function(){
193 $("#dropdownoptions").dialog("open");
194 var target = $(this);
195 $("#dropdownoptions").dialog("widget").position({
196 my: 'left bottom',
197 at: 'left bottom',
198 of: target
Marc Kupietzdab9f222017-11-29 14:22:59 +0100199 });
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100200 });
Marc Kupietzdab9f222017-11-29 14:22:59 +0100201 });
202
Marc Kupietz4abcd682017-11-28 20:51:08 +0100203 $( function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100204 $( "#no_iterations" ).spinner({
205 spin: function( event, ui ) {
206 if ( ui.value < 1000 ) {
207 $( this ).spinner( "value", 1000 );
208 return false;
209 } else if ( ui.value > 10000 ) {
210 $( this ).spinner( "value", 10000 );
211 return false;
212 }
213 }
214 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100215 } );
Marc Kupietz3305b0a2017-11-27 10:46:20 +0100216
Marc Kupietz4abcd682017-11-28 20:51:08 +0100217 $( function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100218 $( "#neighbours" ).spinner({
219 spin: function( event, ui ) {
220 if ( ui.value < 0 ) {
221 $( this ).spinner( "value", 0 );
222 return false;
223 } else if ( ui.value > 200 ) {
224 $( this ).spinner( "value", 200 );
225 return false;
226 }
227 }
228 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100229 } );
230
231 $( function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100232 $( "#cutoff" ).spinner({
233 spin: function( event, ui ) {
234 if ( ui.value < 100000 ) {
235 $( this ).spinner( "value", 100000 );
236 return false;
237 } else if ( ui.value > 2000000 ) {
238 $( this ).spinner( "value", 2000000 );
239 return false;
240 }
241 }
242 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100243 } );
244
245 $( function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100246 $( "#tabs" ).tabs().addClass('tabs-min');
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100247 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100248
249 $( function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100250 $( ".controlgroup-vertical" ).controlgroup({
251 "direction": "vertical"
252 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100253 } );
254
255 $(function() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100256 $( document ).tooltip({
257 content: function() {
258 return $(this).attr('title');
259 }}
260 )
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100261 })
Marc Kupietz694610d2017-11-25 18:30:03 +0100262
Marc Kupietz83305222016-04-28 09:57:22 +0200263 </script>
264 <script src="//d3js.org/d3.v3.min.js" charset="utf-8"></script>
Marc Kupietz554aff52017-11-09 14:42:09 +0100265 <script src="/derekovecs/js/tsne.js"></script>
266 <script src="/derekovecs/js/som.js"></script>
267 <script src="/derekovecs/js/labeler.js"></script>
Marc Kupietz83305222016-04-28 09:57:22 +0200268 <style>
269 body, input {
Marc Kupietz6dbadd12017-11-29 16:43:33 +0100270 font-family: Lato, sans-serif;
Marc Kupietz83305222016-04-28 09:57:22 +0200271 font-size: 11pt;
272 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100273
Marc Kupietz6dbadd12017-11-29 16:43:33 +0100274 h1, h2, h3 {
275 margin: 5px 10px 0 0;
276 color: rgb(246,168,0);
277 font-family: "Univers LT Std 47 Cn Lt", "Univers LT Std 67 Cn Lt", "Roboto Condensed", "Univers LT Std 67 Cn Bold", "UniversLTStd-BoldCn", "Times", 'League Gothic', Impact, sans-serif;
278 font-weight: bold;
279 line-height: 1.35;
280 letter-spacing: normal;
281 text-transform: uppercase;
282 text-shadow: none;
Marc Kupietz34c08172017-11-29 17:08:47 +0100283 word-wrap: break-word;
284 }
285
286
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100287 showoptions, #SEARCH {
288 margin-left: 10px;
289 margin-right: 10px;
Marc Kupietz6dbadd12017-11-29 16:43:33 +0100290 }
291
Marc Kupietzdab9f222017-11-29 14:22:59 +0100292 .tabs-left-vertical .ui-tabs-nav {
293 position: absolute;
294 width: 21em;
295 transform: translate(-100%,0%) rotate(-90deg);
296 transform-origin: 100% 0%;
297 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100298
Marc Kupietzdab9f222017-11-29 14:22:59 +0100299 .tabs-left-vertical .ui-tabs-nav li {
300 float: right;
301 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100302
Marc Kupietzdab9f222017-11-29 14:22:59 +0100303 .tabs-left-vertical .ui-tabs-panel {
304 padding-left: 3.5em;
305 }
306
307 .tabs-left-vertical .ui-tabs-panel {
308 height: 20em;
Marc Kupietz34c08172017-11-29 17:08:47 +0100309 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100310
Marc Kupietz34c08172017-11-29 17:08:47 +0100311 .mono {
312 font-family: "DejaVu Sans Mono", Inconsolata, SourceCodePro, Courier;
313 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100314
Marc Kupietz34c08172017-11-29 17:08:47 +0100315 .ui-tooltip-content {
316 font-size: 9pt;
317 color: #222222;
318 }
Marc Kupietzf4b49392016-04-28 10:49:56 +0200319
Marc Kupietz34c08172017-11-29 17:08:47 +0100320 svg > .ui-tooltip-content {
321 font-size: 8pt;
322 color: #222222;
323 }
324
325 a.merged {
326 color: green;
327 fill: green;
328 }
329
330 #first a {
331 text-decoration: none;
332 }
333
334 a.marked, #first a.marked {
335 text-decoration: underline;
336 }
337
338 a.target {
339 color: red;
340 fill: red;
341 }
Marc Kupietz694610d2017-11-25 18:30:03 +0100342
Marc Kupietz4abcd682017-11-28 20:51:08 +0100343 table.display {
344 width: 40% important!;
345 margin: 1; /* <- works for me this way ****/
346 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100347
Marc Kupietz4abcd682017-11-28 20:51:08 +0100348 table.dataTable thead th, table.dataTable thead td, table.dataTable tbody td {
349 padding: 2px 2px;
350 // border-bottom: 1px solid #111;
351 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100352
Marc Kupietz34c08172017-11-29 17:08:47 +0100353 #collocators {
354 margin-bottom: 15px;
355 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100356
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100357 #header {
358 width: 100%;
359 // border: 1px solid red;
360 overflow: hidden; /* will contain if #first is longer than #second */
361 }
362
Marc Kupietz34c08172017-11-29 17:08:47 +0100363 #topwrapper {
364 width: 100%;
365 // border: 1px solid red;
366 overflow: hidden; /* will contain if #first is longer than #second */
367 }
368
369 #wrapper {
370 // border: 1px solid red;
371 overflow: hidden; /* will contain if #first is longer than #second */
372 }
373
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100374 #pagetitle {
375 max-width: 460px;
376 margin-right: 20px;
377 float: left;
378 overflow: hidden; /* if you don't want #second to wrap below #first */
379 // border: 1px solid green;
380 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100381
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100382 #options {
383 float: left;
384 width: 800px;
385 margin: 10px;
386 overflow: hidden; /* if you don't want #second to wrap below #first */
387 }
388
389 #word {
390 width: 50%;
391 }
392
Marc Kupietz34c08172017-11-29 17:08:47 +0100393 #first {
394 margin-right: 20px;
395 float: left;
396 overflow: hidden; /* if you don't want #second to wrap below #first */
397 // border: 1px solid green;
398 }
399 #tabs {
400 margin-right: 20px;
401 overflow: hidden; /* if you don't want #second to wrap below #first */
402 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100403
404 .tabs-min {
405 background: transparent;
406 border: none;
407 }
408
409 .tabs-min .ui-widget-header {
410 background: transparent;
411 border: none;
412 border-bottom: 1px solid #c0c0c0;
413 -moz-border-radius: 0px;
414 -webkit-border-radius: 0px;
415 border-radius: 0px;
416 }
417
418 .tabs-min .ui-tabs-nav .ui-state-default {
419 background: transparent;
420 border: none;
421 }
422
423 .tabs-min .ui-tabs-nav .ui-state-active {
424 background: transparent url(img/uiTabsArrow.png) no-repeat bottom center;
425 border: none;
426 }
427
428 .tabs-min .ui-tabs-nav .ui-state-default a {
429 color: #c0c0c0;
430 }
431
432 .tabs-min .ui-tabs-nav .ui-state-active a {
433 color: rgb(246,168,0);
434 }
435
Marc Kupietz4abcd682017-11-28 20:51:08 +0100436 #embed {
437 max-width: 802px;
438 border: 1px solid #333;
439 }
440
Marc Kupietz34c08172017-11-29 17:08:47 +0100441 #second {
442 min-width: 800px;
Marc Kupietzdab9f222017-11-29 14:22:59 +0100443 // border: 1px solid #333;
Marc Kupietz34c08172017-11-29 17:08:47 +0100444 overflow: hidden; /* if you don't want #second to wrap below #first */
445 }
446 #som2 svg {
447 border: 1px solid #333;
448 }
Marc Kupietz83305222016-04-28 09:57:22 +0200449
Marc Kupietz34c08172017-11-29 17:08:47 +0100450 #cost {
451 font-size: 8pt;
452 color: #222222;
453 margin-top: 4px;
454 margin-bottom: 12px;
455 }
Marc Kupietz83305222016-04-28 09:57:22 +0200456
Marc Kupietz34c08172017-11-29 17:08:47 +0100457 #sominfo1, #sominfo {
458 font-size: 8pt;
459 color: #222222;
460 margin-top: 0px;
461 }
Marc Kupietz83305222016-04-28 09:57:22 +0200462
Marc Kupietz34c08172017-11-29 17:08:47 +0100463 #somcolor1, #somcolor2, #somcolor3 {
464 display: inline-block;
465 height: 10px;
466 width: 10px;
467 }
Marc Kupietz83305222016-04-28 09:57:22 +0200468
Marc Kupietz34c08172017-11-29 17:08:47 +0100469 #third {
470 border: 1px solid #333;
471 }
Marc Kupietz83305222016-04-28 09:57:22 +0200472
473 </style>
474 <script>
475
476 var opt = {epsilon: <%= $epsilon %>, perplexity: <%= $perplexity %>},
477 mapWidth = 800, // width map
478 mapHeight = 800,
479 jitterRadius = 7;
480
481 var T = new tsnejs.tSNE(opt); // create a tSNE instance
482
483 var Y;
484
485 var data;
486 var labeler;
487
488 function applyJitter() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100489 svg.selectAll('.tsnet')
490 .data(labels)
491 .transition()
492 .duration(50)
493 .attr("transform", function(d, i) {
494 T.Y[i][0] = (d.x - mapWidth/2 - tx)/ss/20;
495 T.Y[i][1] = (d.y - mapHeight/2 - ty)/ss/20;
496 return "translate(" +
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100497 (d.x) + "," +
498 (d.y) + ")";
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100499 });
Marc Kupietz83305222016-04-28 09:57:22 +0200500 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100501
Marc Kupietz83305222016-04-28 09:57:22 +0200502 function updateEmbedding() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100503 var Y = T.getSolution();
504 svg.selectAll('.tsnet')
505 .data(data.words)
506 .attr("transform", function(d, i) {
507 return "translate(" +
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100508 ((Y[i][0]*20*ss + tx) + mapWidth/2) + "," +
509 ((Y[i][1]*20*ss + ty) + mapHeight/2) + ")"; });
Marc Kupietz83305222016-04-28 09:57:22 +0200510 }
511
512 var svg;
513 var labels = [];
514 var anchor_array = [];
515 var text;
516
517 function drawEmbedding() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100518 $("#embed").empty();
519 var div = d3.select("#embed");
Marc Kupietz34c08172017-11-29 17:08:47 +0100520
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100521 // get min and max in each column of Y
522 var Y = T.Y;
Marc Kupietz34c08172017-11-29 17:08:47 +0100523
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100524 svg = div.append("svg") // svg is global
525 .attr("width", mapWidth)
526 .attr("height", mapHeight);
Marc Kupietz34c08172017-11-29 17:08:47 +0100527
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100528 var g = svg.selectAll(".b")
529 .data(data.words)
530 .enter().append("g")
531 .attr("class", "tsnet");
Marc Kupietz34c08172017-11-29 17:08:47 +0100532
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100533 g.append("a")
534 .attr("xlink:href", function(word) {
535 return (data.urlprefix+word);})
536 .attr("class", function(d, i) {
537 var res="";
538 if(data.marked[i]) {
539 res="marked ";
540 }
541 if(data.target.indexOf(" "+d+" ") >= 0) {
542 return res+"target";
543 } else if(data.ranks[i] < data.mergedEnd) {
544 return res+"merged";
545 } else {
546 return res;
547 }
548 })
549 .attr("title", function(d, i) {
550 if(data.mergedEnd > 0) {
551 if(data.ranks[i] >= data.mergedEnd) {
552 return "rank: "+i +" "+"freq. rank: "+(data.ranks[i]).toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
553 } else {
554 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",") + " (merged vocab)";
555 }
556 } else {
557 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
558 }
559 })
560 .append("text")
561 .attr("text-anchor", "top")
562 .attr("font-size", 12)
563 .text(function(d) { return d; });
Marc Kupietz34c08172017-11-29 17:08:47 +0100564
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100565 var zoomListener = d3.behavior.zoom()
566 .scaleExtent([0.1, 10])
567 .center([0,0])
568 .on("zoom", zoomHandler);
569 zoomListener(svg);
Marc Kupietz83305222016-04-28 09:57:22 +0200570 }
571
572 var tx=0, ty=0;
573 var ss=1;
574 var iter_id=-1;
575
576 function zoomHandler() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100577 tx = d3.event.translate[0];
578 ty = d3.event.translate[1];
579 ss = d3.event.scale;
580 updateEmbedding();
Marc Kupietz83305222016-04-28 09:57:22 +0200581 }
582
583 var stepnum = 0;
584
585 function stopStep() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100586 clearInterval(iter_id);
587 text = svg.selectAll("text");
Marc Kupietz34c08172017-11-29 17:08:47 +0100588
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100589 // jitter function needs different data and co-ordinate representation
590 labels = d3.range(data.words.length).map(function(i) {
591 var x = (T.Y[i][0]*20*ss + tx) + mapWidth/2;
592 var y = (T.Y[i][1]*20*ss + ty) + mapHeight/2;
593 anchor_array.push({x: x, y: y, r: jitterRadius});
594 return {
595 x: x,
596 y: y,
597 name: data.words[i]
598 };
599 });
Marc Kupietz34c08172017-11-29 17:08:47 +0100600
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100601 // get the actual label bounding boxes for the jitter function
602 var index = 0;
603 text.each(function() {
604 labels[index].width = this.getBBox().width;
605 labels[index].height = this.getBBox().height;
606 index += 1;
607 });
Marc Kupietz83305222016-04-28 09:57:22 +0200608
Marc Kupietz34c08172017-11-29 17:08:47 +0100609
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100610 // setTimeout(updateEmbedding, 1);
611 // setTimeout(
612 labeler = d3.labeler()
613 .label(labels)
614 .anchor(anchor_array)
615 .width(mapWidth)
616 .height(mapHeight)
617 .update(applyJitter);
618 // .start(1000);
Marc Kupietz83305222016-04-28 09:57:22 +0200619
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100620 iter_id = setInterval(jitterStep, 1);
Marc Kupietz83305222016-04-28 09:57:22 +0200621 }
622
623 var jitter_i=0;
624
625 function jitterStep() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100626 if(jitter_i++ > 100) {
627 clearInterval(iter_id);
628 } else {
629 labeler.start2(10);
630 applyJitter();
631 }
Marc Kupietz83305222016-04-28 09:57:22 +0200632 }
633
634 var last_cost=1000;
635
636 function step() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100637 var i = T.iter;
Marc Kupietz34c08172017-11-29 17:08:47 +0100638
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100639 if(i > <%= $no_iterations %>) {
640 stopStep();
641 } else {
642 var cost = Math.round(T.step() * 100000) / 100000; // do a few steps
643 $("#cost").html("tsne iteration " + i + ", cost: " + cost.toFixed(5));
644 if(i % 250 == 0 && cost >= last_cost) {
645 stopStep();
Marc Kupietz4abcd682017-11-28 20:51:08 +0100646 } else {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100647 last_cost = cost;
648 updateEmbedding();
Marc Kupietz4abcd682017-11-28 20:51:08 +0100649 }
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100650 }
Marc Kupietz83305222016-04-28 09:57:22 +0200651 }
652
653 function showMap(j) {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100654 data=j;
655 T.iter=0;
656 T.initDataRaw(data.vecs); // init embedding
657 drawEmbedding(); // draw initial embedding
Marc Kupietz78114532017-11-29 17:00:16 +0100658
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100659 if(iter_id >= 0) {
660 clearInterval(iter_id);
661 }
662 //T.debugGrad();
663 iter_id = setInterval(step, 1);
664 if(true) { // (<%= $show_som %>) {
665 makeSOM(j, <%= $no_iterations %>);
666 }
Marc Kupietz83305222016-04-28 09:57:22 +0200667 }
Marc Kupietz39179ab2017-07-04 16:28:06 +0200668 var queryword;
669
670 function onload() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100671 queryword = document.getElementById('word');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200672 }
673
674 function queryKorAP() {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100675 window.open('http://korap.ids-mannheim.de/kalamar/?q='+queryword.value, 'KorAP');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200676 }
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100677
678 function queryKorAPCII(query) {
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100679 window.open('http://korap.ids-mannheim.de/kalamar/?ql=cosmas2&q='+query, 'KorAP');
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100680 }
Marc Kupietz83305222016-04-28 09:57:22 +0200681 </script>
682 </head>
Marc Kupietz39179ab2017-07-04 16:28:06 +0200683 <body onload="onload()">
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100684 <div id="header">
685 <div id="pagetitle">
686 <h1>DeReKo-Vectors</h1>
687 </div>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100688 <div id="options" class="widget">
Marc Kupietzdab9f222017-11-29 14:22:59 +0100689 <form id="queryform">
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100690 <input id="word" type="text" name="word" placeholder="Word(s) to be searched" value="<%= $word %>"
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100691 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."/>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100692 <input id="SEARCH" type="button" value="SEARCH">
693 <input type="button" id="showoptions" name="showoptions" value="Options" />
694 </form>
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100695 <div id="dropdownoptions" style="display: none">
Marc Kupietzdab9f222017-11-29 14:22:59 +0100696 <form id="optionsform">
Marc Kupietz4abcd682017-11-28 20:51:08 +0100697 <div class="controlgroup-vertical">
698 <label for="cutoff">cut-off</label>
699 <input id="cutoff" type="text" name="cutoff" size="10" value="<%= $cutoff %>" title="Only consider the most frequent x word forms.">
700 <label for="dedupe">dedupe</label>
701 <input id="dedupe" type="checkbox" name="dedupe" value="1" <%= ($dedupe ? "checked" : "") %> title="radically filter out any near-duplicates">
702 % if($mergedEnd > 0) {
Marc Kupietz78114532017-11-29 17:00:16 +0100703 <label for="sbf">backw.</label>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100704 <input id="sbf" type="checkbox" name="sbf" value="1" <%= ($searchBaseVocabFirst ? "checked" : "") %> title="If checkecked base vocabulary will be searched first. Otherwise merged vocabulray will be searched first.">
705 % }
706 <label for="neighbours">max. neighbours:</label>
707 <input id="neighbours" size="4" name="n" value="<%= $no_nbs %>">
Marc Kupietzc1e42c32017-11-29 16:47:56 +0100708 <label for="no_iterations">max. iterations</label>
709 <input id="no_iterations" name="N" size="4" value="<%= $no_iterations %>">
Marc Kupietz78114532017-11-29 17:00:16 +0100710 <!-- <label for="dosom">SOM</label>
711 <input id="dosom" type="checkbox" name="som" value="1" <%= ($show_som ? "checked" : "") %>> -->
Marc Kupietz4abcd682017-11-28 20:51:08 +0100712 % if($collocators) {
713 <label for="sortby">window/sort</label>
714 <select id="sortby" name="sort">
715 <option value="0" <%= ($sort!=1 && $sort!=2? "selected":"") %>>auto focus</option>
Marc Kupietza77acce2017-11-30 16:59:07 +0100716 <!-- <option value="1" <%= ($sort==1? "selected":"") %>>any single position</option>
717 <option value="2" <%= ($sort==2? "selected":"") %>>whole window</option> -->
Marc Kupietz4abcd682017-11-28 20:51:08 +0100718 </select>
719 % }
Marc Kupietzdab9f222017-11-29 14:22:59 +0100720 <input type="button" value="→ KorAP" onclick="queryKorAP();" title="query word with KorAP"/>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100721 </div>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100722 </form>
723 </div>
Marc Kupietz0af83e32017-11-27 09:31:37 +0100724 </div>
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100725 </div>
726 <div id="topwrapper">
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100727 <div style="visibility: hidden;" id="tabs">
Marc Kupietz4abcd682017-11-28 20:51:08 +0100728 <ul>
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100729 <li><a href="#tabs-1">Semantics (TSNE-map)</a></li>
730 <li><a href="#tabs-2">Semantics (SOM)</a></li>
731 <li><a href="#tabs-3">Syntagmatic (collocators)</a></li>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100732 </ul>
733 <div id="tabs-1">
Marc Kupietzdab9f222017-11-29 14:22:59 +0100734 % if($lists && (@$lists) > 0 && (@$lists)[0]) {
735 <div id="wrapper">
Marc Kupietzaaae0152017-12-01 23:31:56 +0100736 <div id="first" style="width: 320px">
Marc Kupietzdab9f222017-11-29 14:22:59 +0100737 <table class="display compact nowrap" id="firsttable">
738 <thead>
739 <tr>
Marc Kupietzaaae0152017-12-01 23:31:56 +0100740 <th align="right">#</th><th align="right">cos</th><th align="left">similars</th>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100741 </tr>
742 </thead>
743 <tbody>
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100744 <tr>
745 <td align="right">
746 </td>
747 <td align="right">
748 </td>
749 <td></td>
750 </tr>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100751 </tbody>
752 </table>
753 </div>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100754 <script>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100755 </script>
Marc Kupietz896c9092017-12-02 14:40:43 +0100756 <div id="second">
757 <div id="embed">
758 </div>
759 <div id="cost">
760 </div>
761 </div>
762 </div>
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100763 % } elsif($word !~ /^\s*$/) {
764 <div id="wrapper">
765 <div id="not-found-dialog" title="Not found">
766 <p>ERROR: "<%= $word %>" not found in vocabluary.</p>
767 <p>If you are sure you have spelled the word as intended, you can try to increase the cutoff parameter in the options menu.</p>
768 </div>
769 <script>
770 $( function() {
Marc Kupietz896c9092017-12-02 14:40:43 +0100771 $( "#not-found-dialog" ).dialog({
772 autoOpen: true,
773 modal: true,
774 draggable: false,
775 height: "auto",
776 width: "auto",
777 resizable: false,
778 buttons: {
779 "OK": function() {
780 $( this ).dialog( "close" );
781 },
782 "Apply": function() {
783 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100784 }
Marc Kupietz896c9092017-12-02 14:40:43 +0100785 }
Marc Kupietz2f6b74a2017-12-01 13:20:21 +0100786 });
Marc Kupietz896c9092017-12-02 14:40:43 +0100787 });
788 </script>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100789 </div>
790 % }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100791 </div>
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100792 <div id="tabs-2">
793 <div id="som2" style="width: 800; height: 800px">
794 </div>
795 <div id="sominfo1"><span id="somcolor1"> </span> <span id="somword1"> </span> <span id="somcolor2"> </span> <span id="somword2"> </span> <span id="somcolor3"> </span></div>
796 <div id="sominfo">SOM iteration <span id="iterations">0</span></div>
797 </div>
798 <div id="tabs-3">
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100799 <div style="width: 800px" id="secondt">
Marc Kupietz4abcd682017-11-28 20:51:08 +0100800 <table class="display compact nowrap" id="secondtable">
801 <thead>
802 <tr>
803 % if($collocators) {
Marc Kupietzdab9f222017-11-29 14:22:59 +0100804 <th>#</th>
Marc Kupietz22796142017-12-01 13:19:15 +0100805 <th align="center" title="The columns (c) around the target are considered for summation are marked with *.">w'</th>
Marc Kupietzd64f3f22017-11-30 12:07:42 +0100806 <th align="right" title="Raw (max.) activation of the collocator in the output layers.">max(a)</th>
Marc Kupietza77acce2017-11-30 16:59:07 +0100807 <th title="Co-normalized raw activation sum of the collocator in the selected columns." align="right">⊥Σa</th>
808 <th title="Sum of activations over the selected colunns normalized by the total activation sum of the selected columns." align="right">Σa/Σc</th>
809 <th title="Sum of the column normalized activations over the selected colunns." align="right">Σ(a/c)</th>
810 <th title="Sum of the activations over the whole window normalized by the total window sum (no auto-focus)." align="right">Σa/Σw</th>
811 <th align="left">collocator</th>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100812 % }
813 </tr>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100814 </thead>
815 <tbody>
Marc Kupietz22796142017-12-01 13:19:15 +0100816 <tr>
817 <td align="right">
818 </td>
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100819 <td align="right">
820 </td>
821 <td align="right">
822 </td>
823 <td align="right">
824 </td>
825 <td align="right">
826 </td>
827 <td align="right">
828 </td>
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100829<script
830 src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js"
831 </td>
Marc Kupietz22796142017-12-01 13:19:15 +0100832 </tr>
Marc Kupietzdab9f222017-11-29 14:22:59 +0100833 </tbody>
834 </table>
835 </div> <!-- - tab2 -->
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100836 </div> <!-- tabs -->
837 </div>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100838 </div> <!-- topwrapper -->
839 <div style="clear: both;"></div>
Marc Kupietz0af83e32017-11-27 09:31:37 +0100840 </div>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100841 % if($training_args) {
842 <p>
843 Word vector model trained with <a href="https://code.google.com/p/word2vec/">word2vec</a> using the following parameters: <pre><%= $training_args %></pre>
844 </p>
Marc Kupietz83305222016-04-28 09:57:22 +0200845 % }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100846 </body>
Marc Kupietz83305222016-04-28 09:57:22 +0200847</html>