blob: 0a2a408b7bc6edd45d51f50a23a3b62bd2f6269c [file] [log] [blame]
Marc Kupietz83305222016-04-28 09:57:22 +02001<!DOCTYPE html>
2<html>
3 <head>
Marc Kupietz58270662017-12-04 12:10:06 +01004 <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 Kupietz58270662017-12-04 12:10:06 +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 Kupietz58270662017-12-04 12:10:06 +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>
10 <script src = "https://cdn.datatables.net/plug-ins/1.10.16/sorting/scientific.js"></script>
11 <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 Kupietz58270662017-12-04 12:10:06 +010013 src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js"
14 integrity="sha256-VazP97ZCwtekAsvgPBSUwPFKdrwD3unUfSGVYrahUqU="
15 crossorigin="anonymous"></script>
Marc Kupietz80bd7b92017-07-04 16:25:54 +020016 <script>
Marc Kupietz58270662017-12-04 12:10:06 +010017 var urlParams = new URLSearchParams(window.location.search);
Marc Kupietza6e08f02017-12-01 22:06:21 +010018
Marc Kupietz4abcd682017-11-28 20:51:08 +010019 $('#firstable').hide();
Marc Kupietz58270662017-12-04 12:10:06 +010020 $(document).ready(function() {
Marc Kupietz694610d2017-11-25 18:30:03 +010021
Marc Kupietz58270662017-12-04 12:10:06 +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 Kupietz58270662017-12-04 12:10:06 +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 Kupietz58270662017-12-04 12:10:06 +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 Kupietz58270662017-12-04 12:10:06 +010043 function changeCharColor(txt, heat) {
44 var newText = "";
45 for (var i=0, l=txt.length; i<l; i++) {
46 newText += (i == 5 ? txt.charAt(i) : '<span style="background-color:'+getHeatColor(heat[i]/maxHeat)+'">'+txt.charAt(i)+'</span>');
47 }
48 return newText;
Marc Kupietzb6c615d2017-12-02 10:38:20 +010049 }
Marc Kupietzb6c615d2017-12-02 10:38:20 +010050
Marc Kupietz58270662017-12-04 12:10:06 +010051 function getHeatColor(value) {
52 var hue=((1-value)*120).toString(10);
53 return ["hsl(",hue,",90%,70%)"].join("");
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +010054 }
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +010055
Marc Kupietz58270662017-12-04 12:10:06 +010056 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);
Marc Kupietzddaba632017-12-02 17:30:56 +010062 }
Marc Kupietz6e2fc102017-12-01 22:07:23 +010063
Marc Kupietz58270662017-12-04 12:10:06 +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);
Marc Kupietz451f8d82017-12-04 13:19:47 +010068 if (paraResults.length > 0 && paraResults[0] != null) {
Marc Kupietz58270662017-12-04 12:10:06 +010069 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 ],
Marc Kupietzd6b15442017-12-04 13:20:27 +0100101 "oLanguage": {
102 "sSearch": "Filter: "
103 },
Marc Kupietz58270662017-12-04 12:10:06 +0100104 "order": [[ 1, 'desc' ]],
105 } );
Marc Kupietz6e2fc102017-12-01 22:07:23 +0100106
Marc Kupietz58270662017-12-04 12:10:06 +0100107 t.on( 'order.dt search.dt', function () {
108 t.column(0, {order:'applied'}).nodes().each( function (cell, i) {
109 cell.innerHTML = i+1;
110 } );
111 } ).draw();
Marc Kupietzdab9f222017-11-29 14:22:59 +0100112
Marc Kupietz58270662017-12-04 12:10:06 +0100113 $( "#first" ).clone().prependTo( "#tabs-2" );
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100114
Marc Kupietz58270662017-12-04 12:10:06 +0100115 }
Marc Kupietzdab9f222017-11-29 14:22:59 +0100116
Marc Kupietz58270662017-12-04 12:10:06 +0100117 var collocatorData = <%= b(Mojo::JSON::to_json($collocators)) %>;
118 var maxHeat; // = Math.max.apply(Math,collocatorData.map(function(o){return o.cprob;}))
Marc Kupietz3305b0a2017-11-27 10:46:20 +0100119
Marc Kupietz58270662017-12-04 12:10:06 +0100120 if (collocatorData != null) {
121 maxHeat = Math.max.apply(Math,collocatorData.map(function(o){return Math.max.apply(Math,o.heat);}))
122 var t = $('#secondtable').DataTable({
123 data: collocatorData,
124 "sScrollY": "780px",
125 "bScrollCollapse": true,
126 "bPaginate": false,
127 "bJQueryUI": true,
128 "dom": '<"top">rt<"bottom"flp><"clear">',
129 "columns": [
130 { "data": "rank", type: "allnumeric" },
131 { "data": "pos", width: "7%", sClass: "dt-center mono compact", render: function ( data, type, row ) {return bitvec2window(data, row.heat) }},
132 { "data": "max", render: function ( data, type, row ) {return data.toFixed(3) }},
133 { "data": "conorm", render: function ( data, type, row ) {return data.toFixed(3) }},
134 { "data": "prob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
135 { "data": "cprob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
136 { "data": "overall", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
137 { "data": "word", sClass: "collocator" }
138 ],
139 "columnDefs": [
140 { className: "dt-right", "targets": [0,2,3,4,5,6] },
141 { className: "dt-center", "targets": [ 1] },
142 { "searchable": false,
143 "orderable": false,
144 "targets": 0
145 },
146 { "type": "scientific", targets: [2,3,4,5,6] },
147 { "orderSequence": [ "desc" ], "targets": [ 2, 3, 4, 5, 6 ] },
148 { "orderSequence": [ "asc", "desc" ], "targets": [ 1, 7 ] },
149 ],
Marc Kupietzd6b15442017-12-04 13:20:27 +0100150 "oLanguage": {
151 "sSearch": "Filter: "
152 },
Marc Kupietz58270662017-12-04 12:10:06 +0100153 "order": [[ 4, 'desc' ]],
154 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100155
Marc Kupietz58270662017-12-04 12:10:06 +0100156 t.on( 'order.dt search.dt', function () {
157 t.column(0, {order:'applied'}).nodes().each( function (cell, i) {
158 cell.innerHTML = i+1;
159 } );
160 } ).draw();
161
162 if (localStorage) { // let's not crash if some user has IE7
163 var index = parseInt(localStorage['tab']||'0');
164 $("#tabs").tabs({ active: index });
165 }
166 }
167 $("#tabs").css("visibility", "visible"); // now we can show the tabs
168
169 });
170
171 $(function(){
172 $("#dropdownoptions").dialog({
173 title: "Options",
174 autoOpen: false,
175 modal: false,
176 draggable: false,
177 height: "auto",
178 width: "auto",
179 resizable: false,
180 buttons: {
181 "Cancel": function() {
182 $( this ).dialog( "close" );
183 },
184 "Apply": function() {
185 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
186 }
187 }
188 });
189 });
190
191 $(function(){
192 $("td.collocator").click(function(){
193 queryKorAPCII(this.textContent + " /w5 " + urlParams.get('word'));
194 });
195 });
196
197 $(function(){
198 $("#showoptions").click(function(){
199 $("#dropdownoptions").dialog("open");
200 var target = $(this);
201 $("#dropdownoptions").dialog("widget").position({
202 my: 'left bottom',
203 at: 'left bottom',
204 of: target
205 });
206 });
207 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100208
209 $( function() {
Marc Kupietz58270662017-12-04 12:10:06 +0100210 $( "#no_iterations" ).spinner({
211 spin: function( event, ui ) {
212 if ( ui.value < 1000 ) {
213 $( this ).spinner( "value", 1000 );
214 return false;
215 } else if ( ui.value > 10000 ) {
216 $( this ).spinner( "value", 10000 );
217 return false;
218 }
219 }
220 });
221 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100222
Marc Kupietz58270662017-12-04 12:10:06 +0100223 $( function() {
224 $( "#neighbours" ).spinner({
225 spin: function( event, ui ) {
226 if ( ui.value < 0 ) {
227 $( this ).spinner( "value", 0 );
228 return false;
229 } else if ( ui.value > 200 ) {
230 $( this ).spinner( "value", 200 );
231 return false;
232 }
233 }
234 });
235 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100236
Marc Kupietz58270662017-12-04 12:10:06 +0100237 $( function() {
238 $( "#cutoff" ).spinner({
239 spin: function( event, ui ) {
240 if ( ui.value < 100000 ) {
241 $( this ).spinner( "value", 100000 );
242 return false;
243 } else if ( ui.value > 2000000 ) {
244 $( this ).spinner( "value", 2000000 );
245 return false;
246 }
247 }
248 });
249 } );
250
251 $( function() {
252 $( "#tabs" ).tabs().addClass('tabs-min');
253 } );
254
255 $( function() {
256 $( ".controlgroup-vertical" ).controlgroup({
257 "direction": "vertical"
258 });
259 } );
260
261 $(function() {
262 $( document ).tooltip({
263 content: function() {
264 return $(this).attr('title');
265 }}
266 )
267 })
Marc Kupietz694610d2017-11-25 18:30:03 +0100268
Marc Kupietz83305222016-04-28 09:57:22 +0200269 </script>
Marc Kupietz58270662017-12-04 12:10:06 +0100270 <script src="//d3js.org/d3.v3.min.js" charset="utf-8"></script>
271 <script src="/derekovecs/js/tsne.js"></script>
272 <script src="/derekovecs/js/som.js"></script>
273 <script src="/derekovecs/js/labeler.js"></script>
Marc Kupietz83305222016-04-28 09:57:22 +0200274 <style>
275 body, input {
Marc Kupietz58270662017-12-04 12:10:06 +0100276 font-family: Lato, sans-serif;
277 font-size: 11pt;
Marc Kupietz83305222016-04-28 09:57:22 +0200278 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100279
Marc Kupietz58270662017-12-04 12:10:06 +0100280 h1, h2, h3 {
281 margin: 5px 10px 0 0;
282 color: rgb(246,168,0);
283 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;
284 font-weight: bold;
285 line-height: 1.35;
286 letter-spacing: normal;
287 text-transform: uppercase;
288 text-shadow: none;
289 word-wrap: break-word;
290 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100291
292
Marc Kupietz58270662017-12-04 12:10:06 +0100293 showoptions, #SEARCH {
294 margin-left: 10px;
295 margin-right: 10px;
296 }
Marc Kupietz6dbadd12017-11-29 16:43:33 +0100297
Marc Kupietz58270662017-12-04 12:10:06 +0100298 .tabs-left-vertical .ui-tabs-nav {
299 position: absolute;
300 width: 21em;
301 transform: translate(-100%,0%) rotate(-90deg);
302 transform-origin: 100% 0%;
303 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100304
Marc Kupietz58270662017-12-04 12:10:06 +0100305 .tabs-left-vertical .ui-tabs-nav li {
306 float: right;
307 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100308
Marc Kupietz58270662017-12-04 12:10:06 +0100309 .tabs-left-vertical .ui-tabs-panel {
310 padding-left: 3.5em;
311 }
Marc Kupietzdab9f222017-11-29 14:22:59 +0100312
Marc Kupietz58270662017-12-04 12:10:06 +0100313 .tabs-left-vertical .ui-tabs-panel {
314 height: 20em;
315 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100316
Marc Kupietz58270662017-12-04 12:10:06 +0100317 .mono {
318 font-family: "DejaVu Sans Mono", Inconsolata, SourceCodePro, Courier;
319 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100320
Marc Kupietz58270662017-12-04 12:10:06 +0100321 .ui-tooltip-content {
322 font-size: 9pt;
323 color: #222222;
324 }
Marc Kupietzf4b49392016-04-28 10:49:56 +0200325
Marc Kupietz58270662017-12-04 12:10:06 +0100326 svg > .ui-tooltip-content {
327 font-size: 8pt;
328 color: #222222;
329 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100330
Marc Kupietz58270662017-12-04 12:10:06 +0100331 a.merged {
332 color: green;
333 fill: green;
334 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100335
Marc Kupietz58270662017-12-04 12:10:06 +0100336 #first a {
337 text-decoration: none;
338 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100339
Marc Kupietz58270662017-12-04 12:10:06 +0100340 a.marked, #first a.marked {
341 text-decoration: underline;
342 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100343
Marc Kupietz58270662017-12-04 12:10:06 +0100344 a.target {
345 color: red;
346 fill: red;
347 }
Marc Kupietz694610d2017-11-25 18:30:03 +0100348
Marc Kupietz58270662017-12-04 12:10:06 +0100349 table.display {
350 width: 40% important!;
351 margin: 1; /* <- works for me this way ****/
352 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100353
Marc Kupietz58270662017-12-04 12:10:06 +0100354 table.dataTable thead th, table.dataTable thead td, table.dataTable tbody td {
355 padding: 2px 2px;
356 // border-bottom: 1px solid #111;
357 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100358
Marc Kupietz58270662017-12-04 12:10:06 +0100359 #collocators {
360 margin-bottom: 15px;
361 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100362
Marc Kupietz58270662017-12-04 12:10:06 +0100363 #header {
364 width: 100%;
365 // border: 1px solid red;
366 overflow: hidden; /* will contain if #first is longer than #second */
367 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100368
Marc Kupietz58270662017-12-04 12:10:06 +0100369 #topwrapper {
370 width: 100%;
371 // border: 1px solid red;
372 overflow: hidden; /* will contain if #first is longer than #second */
373 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100374
Marc Kupietz58270662017-12-04 12:10:06 +0100375 #wrapper {
376 // border: 1px solid red;
377 overflow: hidden; /* will contain if #first is longer than #second */
378 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100379
Marc Kupietz58270662017-12-04 12:10:06 +0100380 #pagetitle {
381 max-width: 460px;
382 margin-right: 20px;
383 float: left;
384 overflow: hidden; /* if you don't want #second to wrap below #first */
385 // border: 1px solid green;
386 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100387
Marc Kupietz58270662017-12-04 12:10:06 +0100388 #options {
389 float: left;
390 width: 800px;
391 margin: 10px;
392 overflow: hidden; /* if you don't want #second to wrap below #first */
393 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100394
Marc Kupietz58270662017-12-04 12:10:06 +0100395 #word {
396 width: 50%;
397 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100398
Marc Kupietz58270662017-12-04 12:10:06 +0100399 #first {
400 margin-right: 20px;
401 float: left;
402 overflow: hidden; /* if you don't want #second to wrap below #first */
403 // border: 1px solid green;
404 }
405 #tabs {
406 margin-right: 20px;
407 overflow: hidden; /* if you don't want #second to wrap below #first */
408 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100409
Marc Kupietz58270662017-12-04 12:10:06 +0100410 .tabs-min {
411 background: transparent;
412 border: none;
413 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100414
Marc Kupietz58270662017-12-04 12:10:06 +0100415 .tabs-min .ui-widget-header {
416 background: transparent;
417 border: none;
418 border-bottom: 1px solid #c0c0c0;
419 -moz-border-radius: 0px;
420 -webkit-border-radius: 0px;
421 border-radius: 0px;
422 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100423
Marc Kupietz58270662017-12-04 12:10:06 +0100424 .tabs-min .ui-tabs-nav .ui-state-default {
425 background: transparent;
426 border: none;
427 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100428
Marc Kupietz58270662017-12-04 12:10:06 +0100429 .tabs-min .ui-tabs-nav .ui-state-active {
430 background: transparent url(derekovecs/img/uiTabsArrow.png) no-repeat bottom center;
431 border: none;
432 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100433
Marc Kupietz58270662017-12-04 12:10:06 +0100434 .tabs-min .ui-tabs-nav .ui-state-default a {
435 color: #c0c0c0;
436 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100437
Marc Kupietz58270662017-12-04 12:10:06 +0100438 .tabs-min .ui-tabs-nav .ui-state-active a {
439 color: rgb(246,168,0);
440 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100441
Marc Kupietz58270662017-12-04 12:10:06 +0100442 #embed {
443 max-width: 802px;
444 border: 1px solid #333;
445 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100446
Marc Kupietz58270662017-12-04 12:10:06 +0100447 #second {
448 min-width: 800px;
449 // border: 1px solid #333;
450 overflow: hidden; /* if you don't want #second to wrap below #first */
451 }
452 #som2 svg {
453 border: 1px solid #333;
454 }
Marc Kupietz83305222016-04-28 09:57:22 +0200455
Marc Kupietz58270662017-12-04 12:10:06 +0100456 #cost {
457 font-size: 8pt;
458 color: #222222;
459 margin-top: 4px;
460 margin-bottom: 12px;
461 }
Marc Kupietz83305222016-04-28 09:57:22 +0200462
Marc Kupietz58270662017-12-04 12:10:06 +0100463 #sominfo1, #sominfo {
464 font-size: 8pt;
465 color: #222222;
466 margin-top: 0px;
467 }
Marc Kupietz83305222016-04-28 09:57:22 +0200468
Marc Kupietz58270662017-12-04 12:10:06 +0100469 #somcolor1, #somcolor2, #somcolor3 {
470 display: inline-block;
471 height: 10px;
472 width: 10px;
473 }
Marc Kupietz83305222016-04-28 09:57:22 +0200474
Marc Kupietz58270662017-12-04 12:10:06 +0100475 #third {
476 border: 1px solid #333;
477 }
Marc Kupietz83305222016-04-28 09:57:22 +0200478
479 </style>
480 <script>
481
482 var opt = {epsilon: <%= $epsilon %>, perplexity: <%= $perplexity %>},
Marc Kupietz58270662017-12-04 12:10:06 +0100483 mapWidth = 800, // width map
484 mapHeight = 800,
485 jitterRadius = 7;
Marc Kupietz83305222016-04-28 09:57:22 +0200486
487 var T = new tsnejs.tSNE(opt); // create a tSNE instance
488
489 var Y;
490
491 var data;
492 var labeler;
493
494 function applyJitter() {
Marc Kupietz58270662017-12-04 12:10:06 +0100495 svg.selectAll('.tsnet')
496 .data(labels)
497 .transition()
498 .duration(50)
499 .attr("transform", function(d, i) {
500 T.Y[i][0] = (d.x - mapWidth/2 - tx)/ss/20;
501 T.Y[i][1] = (d.y - mapHeight/2 - ty)/ss/20;
502 return "translate(" +
503 (d.x) + "," +
504 (d.y) + ")";
505 });
Marc Kupietz83305222016-04-28 09:57:22 +0200506 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100507
Marc Kupietz83305222016-04-28 09:57:22 +0200508 function updateEmbedding() {
Marc Kupietz58270662017-12-04 12:10:06 +0100509 var Y = T.getSolution();
510 svg.selectAll('.tsnet')
511 .data(data.words)
512 .attr("transform", function(d, i) {
513 return "translate(" +
514 ((Y[i][0]*20*ss + tx) + mapWidth/2) + "," +
515 ((Y[i][1]*20*ss + ty) + mapHeight/2) + ")"; });
Marc Kupietz83305222016-04-28 09:57:22 +0200516 }
517
518 var svg;
519 var labels = [];
520 var anchor_array = [];
521 var text;
522
523 function drawEmbedding() {
Marc Kupietz58270662017-12-04 12:10:06 +0100524 var urlprefix = new URLSearchParams(window.location.search);
525 $("#embed").empty();
526 var div = d3.select("#embed");
Marc Kupietz34c08172017-11-29 17:08:47 +0100527
Marc Kupietz58270662017-12-04 12:10:06 +0100528 // get min and max in each column of Y
529 var Y = T.Y;
Marc Kupietz34c08172017-11-29 17:08:47 +0100530
Marc Kupietz58270662017-12-04 12:10:06 +0100531 svg = div.append("svg") // svg is global
532 .attr("width", mapWidth)
533 .attr("height", mapHeight);
Marc Kupietz34c08172017-11-29 17:08:47 +0100534
Marc Kupietz58270662017-12-04 12:10:06 +0100535 var g = svg.selectAll(".b")
536 .data(data.words)
537 .enter().append("g")
538 .attr("class", "tsnet");
Marc Kupietz34c08172017-11-29 17:08:47 +0100539
Marc Kupietz58270662017-12-04 12:10:06 +0100540 g.append("a")
541 .attr("xlink:href", function(word) {
542 urlprefix.set("word", word); return "?"+urlprefix; })
543 .attr("class", function(d, i) {
544 var res="";
545 if(data.marked[i]) {
546 res="marked ";
547 }
548 if(data.target.indexOf(" "+d+" ") >= 0) {
549 return res+"target";
550 } else if(data.ranks[i] < data.mergedEnd) {
551 return res+"merged";
552 } else {
553 return res;
554 }
555 })
556 .attr("title", function(d, i) {
557 if(data.mergedEnd > 0) {
558 if(data.ranks[i] >= data.mergedEnd) {
559 return "rank: "+i +" "+"freq. rank: "+(data.ranks[i]).toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
560 } else {
561 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",") + " (merged vocab)";
562 }
563 } else {
564 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
565 }
566 })
567 .append("text")
568 .attr("text-anchor", "top")
569 .attr("font-size", 12)
570 .text(function(d) { return d; });
Marc Kupietz34c08172017-11-29 17:08:47 +0100571
Marc Kupietz58270662017-12-04 12:10:06 +0100572 var zoomListener = d3.behavior.zoom()
573 .scaleExtent([0.1, 10])
574 .center([0,0])
575 .on("zoom", zoomHandler);
576 zoomListener(svg);
Marc Kupietz83305222016-04-28 09:57:22 +0200577 }
578
579 var tx=0, ty=0;
580 var ss=1;
581 var iter_id=-1;
582
583 function zoomHandler() {
Marc Kupietz58270662017-12-04 12:10:06 +0100584 tx = d3.event.translate[0];
585 ty = d3.event.translate[1];
586 ss = d3.event.scale;
587 updateEmbedding();
Marc Kupietz83305222016-04-28 09:57:22 +0200588 }
589
590 var stepnum = 0;
591
592 function stopStep() {
Marc Kupietz58270662017-12-04 12:10:06 +0100593 clearInterval(iter_id);
594 text = svg.selectAll("text");
Marc Kupietz34c08172017-11-29 17:08:47 +0100595
Marc Kupietz58270662017-12-04 12:10:06 +0100596 // jitter function needs different data and co-ordinate representation
597 labels = d3.range(data.words.length).map(function(i) {
598 var x = (T.Y[i][0]*20*ss + tx) + mapWidth/2;
599 var y = (T.Y[i][1]*20*ss + ty) + mapHeight/2;
600 anchor_array.push({x: x, y: y, r: jitterRadius});
601 return {
602 x: x,
603 y: y,
604 name: data.words[i]
605 };
606 });
Marc Kupietz34c08172017-11-29 17:08:47 +0100607
Marc Kupietz58270662017-12-04 12:10:06 +0100608 // get the actual label bounding boxes for the jitter function
609 var index = 0;
610 text.each(function() {
611 labels[index].width = this.getBBox().width;
612 labels[index].height = this.getBBox().height;
613 index += 1;
614 });
Marc Kupietz83305222016-04-28 09:57:22 +0200615
Marc Kupietz34c08172017-11-29 17:08:47 +0100616
Marc Kupietz58270662017-12-04 12:10:06 +0100617 // setTimeout(updateEmbedding, 1);
618 // setTimeout(
619 labeler = d3.labeler()
620 .label(labels)
621 .anchor(anchor_array)
622 .width(mapWidth)
623 .height(mapHeight)
624 .update(applyJitter);
625 // .start(1000);
Marc Kupietz83305222016-04-28 09:57:22 +0200626
Marc Kupietz58270662017-12-04 12:10:06 +0100627 iter_id = setInterval(jitterStep, 1);
Marc Kupietz83305222016-04-28 09:57:22 +0200628 }
629
630 var jitter_i=0;
631
632 function jitterStep() {
Marc Kupietz58270662017-12-04 12:10:06 +0100633 if(jitter_i++ > 100) {
634 clearInterval(iter_id);
635 } else {
636 labeler.start2(10);
637 applyJitter();
638 }
Marc Kupietz83305222016-04-28 09:57:22 +0200639 }
640
641 var last_cost=1000;
642
643 function step() {
Marc Kupietz58270662017-12-04 12:10:06 +0100644 var i = T.iter;
Marc Kupietz34c08172017-11-29 17:08:47 +0100645
Marc Kupietz58270662017-12-04 12:10:06 +0100646 if(i > <%= $no_iterations %>) {
647 stopStep();
648 } else {
649 var cost = Math.round(T.step() * 100000) / 100000; // do a few steps
650 $("#cost").html("tsne iteration " + i + ", cost: " + cost.toFixed(5));
651 if(i % 250 == 0 && cost >= last_cost) {
652 stopStep();
653 } else {
654 last_cost = cost;
655 updateEmbedding();
656 }
657 }
Marc Kupietz83305222016-04-28 09:57:22 +0200658 }
659
660 function showMap(j) {
Marc Kupietz58270662017-12-04 12:10:06 +0100661 data=j;
662 T.iter=0;
663 T.initDataRaw(data.vecs); // init embedding
664 drawEmbedding(); // draw initial embedding
Marc Kupietz78114532017-11-29 17:00:16 +0100665
Marc Kupietz58270662017-12-04 12:10:06 +0100666 if(iter_id >= 0) {
667 clearInterval(iter_id);
668 }
669 //T.debugGrad();
670 iter_id = setInterval(step, 1);
671 if(true) { // (<%= $show_som %>) {
672 makeSOM(j, <%= $no_iterations %>);
673 }
Marc Kupietz83305222016-04-28 09:57:22 +0200674 }
Marc Kupietz39179ab2017-07-04 16:28:06 +0200675 var queryword;
676
677 function onload() {
Marc Kupietz58270662017-12-04 12:10:06 +0100678 queryword = document.getElementById('word');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200679 }
680
681 function queryKorAP() {
Marc Kupietz58270662017-12-04 12:10:06 +0100682 window.open('http://korap.ids-mannheim.de/kalamar/?q='+queryword.value, 'KorAP');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200683 }
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100684
685 function queryKorAPCII(query) {
Marc Kupietz58270662017-12-04 12:10:06 +0100686 window.open('http://korap.ids-mannheim.de/kalamar/?ql=cosmas2&q='+query, 'KorAP');
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100687 }
Marc Kupietz83305222016-04-28 09:57:22 +0200688 </script>
689 </head>
Marc Kupietz39179ab2017-07-04 16:28:06 +0200690 <body onload="onload()">
Marc Kupietz58270662017-12-04 12:10:06 +0100691 <div id="header">
692 <div id="pagetitle">
693 <h1>DeReKo-Vectors</h1>
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100694 </div>
Marc Kupietz58270662017-12-04 12:10:06 +0100695 <div id="options" class="widget">
696 <form id="queryform">
697 <input id="word" type="text" name="word" placeholder="Word(s) to be searched" value="<%= $word %>"
698 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."/>
699 <input id="SEARCH" type="button" value="SEARCH">
700 <input type="button" id="showoptions" name="showoptions" value="Options" />
701 </form>
702 <div id="dropdownoptions" style="display: none">
703 <form id="optionsform">
704 <div class="controlgroup-vertical">
705 <label for="cutoff">cut-off</label>
706 <input id="cutoff" type="text" name="cutoff" size="10" value="<%= $cutoff %>" title="Only consider the most frequent x word forms.">
707 <label for="dedupe">dedupe</label>
708 <input id="dedupe" type="checkbox" name="dedupe" value="1" <%= ($dedupe ? "checked" : "") %> title="radically filter out any near-duplicates">
709 % if($mergedEnd > 0) {
710 <label for="sbf">backw.</label>
711 <input id="sbf" type="checkbox" name="sbf" value="1" <%= ($searchBaseVocabFirst ? "checked" : "") %> title="If checkecked base vocabulary will be searched first. Otherwise merged vocabulray will be searched first.">
712 % }
713 <label for="neighbours">max. neighbours:</label>
714 <input id="neighbours" size="4" name="n" value="<%= $no_nbs %>">
715 <label for="no_iterations">max. iterations</label>
716 <input id="no_iterations" name="N" size="4" value="<%= $no_iterations %>">
717 <!-- <label for="dosom">SOM</label>
718 <input id="dosom" type="checkbox" name="som" value="1" <%= ($show_som ? "checked" : "") %>> -->
719 % if($collocators) {
720 <label for="sortby">window/sort</label>
721 <select id="sortby" name="sort">
722 <option value="0" <%= ($sort!=1 && $sort!=2? "selected":"") %>>auto focus</option>
723 <!-- <option value="1" <%= ($sort==1? "selected":"") %>>any single position</option>
724 <option value="2" <%= ($sort==2? "selected":"") %>>whole window</option> -->
725 </select>
726 % }
727 <input type="button" value="→ KorAP" onclick="queryKorAP();" title="query word with KorAP"/>
728 </div>
729 </form>
730 </div>
731 </div>
732 </div>
733 <div id="topwrapper">
734 <div style="visibility: hidden;" id="tabs">
735 <ul>
736 <li><a href="#tabs-1">Semantics (TSNE-map)</a></li>
737 <li><a href="#tabs-2">Semantics (SOM)</a></li>
738 <li><a href="#tabs-3">Syntagmatic (collocators)</a></li>
739 </ul>
740 <div id="tabs-1">
741 % if($lists && (@$lists) > 0 && (@$lists)[0]) {
742 <div id="wrapper">
743 <div id="first" style="width: 320px">
744 <table class="display compact nowrap" id="firsttable">
745 <thead>
746 <tr>
747 <th align="right">#</th><th align="right">cos</th><th align="left">similars</th>
748 </tr>
749 </thead>
750 <tbody>
751 <tr>
752 <td align="right">
753 </td>
754 <td align="right">
755 </td>
756 <td></td>
757 </tr>
758 </tbody>
759 </table>
760 </div>
761 <script>
762 </script>
763 <div id="second">
764 <div id="embed">
765 </div>
766 <div id="cost">
767 </div>
768 </div>
769 </div>
770 % } elsif($word !~ /^\s*$/) {
771 <div id="wrapper">
772 <div id="not-found-dialog" title="Not found">
773 <p>ERROR: "<%= $word %>" not found in vocabluary.</p>
774 <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>
775 </div>
776 <script>
777 $( function() {
778 $( "#not-found-dialog" ).dialog({
779 autoOpen: true,
780 modal: true,
781 draggable: false,
782 height: "auto",
783 width: "auto",
784 resizable: false,
785 buttons: {
786 "OK": function() {
787 $( this ).dialog( "close" );
788 },
789 "Apply": function() {
790 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
791 }
792 }
793 });
794 });
795 </script>
796 </div>
797 % }
798 </div>
799 <div id="tabs-2">
800 <div id="som2" style="width: 800; height: 800px">
801 </div>
802 <div id="sominfo1"><span id="somcolor1"> </span> <span id="somword1"> </span> <span id="somcolor2"> </span> <span id="somword2"> </span> <span id="somcolor3"> </span></div>
803 <div id="sominfo">SOM iteration <span id="iterations">0</span></div>
804 </div>
805 <div id="tabs-3">
806 <div style="width: 800px" id="secondt">
807 <table class="display compact nowrap" id="secondtable">
808 <thead>
809 <tr>
810 % if($collocators) {
811 <th>#</th>
812 <th align="center" title="The columns (c) around the target are considered for summation are marked with *.">w'</th>
813 <th align="right" title="Raw (max.) activation of the collocator in the output layers.">max(a)</th>
814 <th title="Co-normalized raw activation sum of the collocator in the selected columns." align="right">⊥Σa</th>
815 <th title="Sum of activations over the selected colunns normalized by the total activation sum of the selected columns." align="right">Σa/Σc</th>
816 <th title="Sum of the column normalized activations over the selected colunns." align="right">Σ(a/c)</th>
817 <th title="Sum of the activations over the whole window normalized by the total window sum (no auto-focus)." align="right">Σa/Σw</th>
818 <th align="left">collocator</th>
819 % }
820 </tr>
821 </thead>
822 <tbody>
823 <tr>
824 <td align="right">
825 </td>
826 <td align="right">
827 </td>
828 <td align="right">
829 </td>
830 <td align="right">
831 </td>
832 <td align="right">
833 </td>
834 <td align="right">
835 </td>
836 <script
837 src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js"
838 </td>
839 </tr>
840 </tbody>
841 </table>
842 </div> <!-- - tab2 -->
843 </div> <!-- tabs -->
844 </div>
845 </div> <!-- topwrapper -->
846 <div style="clear: both;"></div>
847 </div>
Marc Kupietz4abcd682017-11-28 20:51:08 +0100848 % if($training_args) {
849 <p>
850 Word vector model trained with <a href="https://code.google.com/p/word2vec/">word2vec</a> using the following parameters: <pre><%= $training_args %></pre>
851 </p>
Marc Kupietz83305222016-04-28 09:57:22 +0200852 % }
Marc Kupietz58270662017-12-04 12:10:06 +0100853 </body>
Marc Kupietz83305222016-04-28 09:57:22 +0200854</html>