Gitiles
Code Review Sign In
korap.ids-mannheim.de / ids-kl / derekovecs / d5d0573442a7d9feb9ee9cc25c55ae5aa4479ec3 / . / templates / index.html.ep
blob: cb833ea323cf51df8e7345614e021e4e8c8ddce3 [file] [log] [blame]
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]1<!DOCTYPE html>
2<html>
3 <head>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]4 <title>DeReKo-Word-Vector-Distances: <%= $word %></title>
Marc Kupietz80bd7b92017-07-04 16:25:54 +0200[diff] [blame]5 <link rel="stylesheet" href="//code.jquery.com/ui/1.12.1/themes/base/jquery-ui.css">
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]6 <link href="https://fonts.googleapis.com/css?family=Lato|Roboto+Condensed" rel="stylesheet">
Marc Kupietz80bd7b92017-07-04 16:25:54 +0200[diff] [blame]7 <script src="http://code.jquery.com/jquery-latest.min.js"></script>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]8 <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>
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]11 <script src='https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.2/MathJax.js?config=TeX-MML-AM_CHTML'></script>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]12 <link rel="stylesheet" href="https://cdn.datatables.net/1.10.16/css/jquery.dataTables.min.css">
Marc Kupietz80bd7b92017-07-04 16:25:54 +0200[diff] [blame]13 <script
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]14 src="http://code.jquery.com/ui/1.12.1/jquery-ui.min.js"
15 integrity="sha256-VazP97ZCwtekAsvgPBSUwPFKdrwD3unUfSGVYrahUqU="
16 crossorigin="anonymous"></script>
Marc Kupietz80bd7b92017-07-04 16:25:54 +0200[diff] [blame]17 <script>
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]18 MathJax.Hub.Config({
19 config: ["MMLorHTML.js"],
20 jax: ["input/TeX","input/MathML","output/HTML-CSS","output/NativeMML", "output/PreviewHTML"],
21 extensions: ["tex2jax.js","mml2jax.js","MathMenu.js","MathZoom.js", "fast-preview.js", "AssistiveMML.js", "a11y/accessibility-menu.js"],
22 TeX: {
23 extensions: ["AMSmath.js","AMSsymbols.js","noErrors.js","noUndefined.js"]
24 }
25 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]26 var urlParams = new URLSearchParams(window.location.search);
Marc Kupietzb3a2e4f2017-12-08 17:25:53 +0100[diff] [blame]27 var currentWords = urlParams.get("word");
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]28 var CIIsearchWords = (currentWords && currentWords.includes(" ") ? '('+currentWords.replace(/ +/g, " oder ")+')' : currentWords);
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]29 var collocatorTable = null;
Marc Kupietza6e08f02017-12-01 22:06:21 +0100[diff] [blame]30
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]31 $(document).ready(function() {
Marc Kupietz694610d2017-11-25 18:30:03 +0100[diff] [blame]32
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]33 $('#firstable').hide();
34 //Set up a callback to hear back when MathJax is done rendering the equations
35 // it finds
36 $('#ccd').load(
37 '@Url.Action("ActionResultMethod","ControllerName",{controller parameters})',
38 function () {
39 MathJax.Hub.Queue(
40 ["Typeset",MathJax.Hub,"ccd"],
41 function () {
42 $("#lfmd_tt").attr("title",$("#lfmd_ttt").html());
43 $("#npmi_tt").attr("title",$("#npmi_ttt").html());
44 }
45 );
46 });
47
48 //set things up so that we can shove raw html into what is shown in the tooltip;
49 // in this case, we will have already put into the title attribute the html that
50 // contains the MathJax rendered equations (via what we do in the callback).
51 $(function () {
52 $(document).tooltip({
53 content: function () {
54 return $(this).prop('title');
55 }
56 });
57 });
58
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]59 $("input").bind("keydown", function(event) {
60 // track enter key
61 var keycode = (event.keyCode ? event.keyCode : (event.which ? event.which : event.charCode));
62 if (keycode == 13) { // keycode for enter key
63 // force the 'Enter Key' to implicitly click the Update button
64 document.getElementById('SEARCH').click();
65 return false;
66 } else {
67 return true;
68 }});
Marc Kupietzdab9f222017-11-29 14:22:59 +0100[diff] [blame]69
Marc Kupietze871abd2018-01-25 16:18:27 +0100[diff] [blame]70 var collocatorTable_activated = false;
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]71 $( "#tabs" ).on( "tabsactivate", function( event, ui ) {
72 if (localStorage) localStorage['tab'] = ui.newTab.index();
Marc Kupietze871abd2018-01-25 16:18:27 +0100[diff] [blame]73 if(ui.newTab.index() == 2 && !collocatorTable_activated) {
74 classicCollocatorTable.columns.adjust();
75 collocatorTable.columns.adjust();
76 collocatorTable_activated = true;
77 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]78 });
79
80 $(function(){
81 $("#SEARCH").click(function() {
82 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]83 });
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]84 });
Marc Kupietz0af83e32017-11-27 09:31:37 +0100[diff] [blame]85
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]86 function changeCharColor(txt, heat, word) {
87 var newText = "";
88 for (var i=0, l=txt.length; i<l; i++) {
89 newText += (i == 5 ? txt.charAt(i) : '<a href="http://korap.ids-mannheim.de/kalamar/?ql=cosmas2&q=' +
90 CIIsearchWords + ' /' + (i > 5? '%2B' : '-') + 'w' +
91 Math.abs(i-5) + ':' + Math.abs(i-5) + ' ' + word +
92 '" target="korap"><span style="background-color:' +
93 getHeatColor(heat[i]/maxHeat)+'">'+txt.charAt(i)+'</span></a>');
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100[diff] [blame]94 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]95 return newText;
96 }
Marc Kupietzb6c615d2017-12-02 10:38:20 +0100[diff] [blame]97
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]98 function getHeatColor(value) {
99 var hue=((1-value)*120).toString(10);
100 return ["hsl(",hue,",90%,70%)"].join("");
101 }
102
103 function bitvec2window(n, heat, word) {
104 var str = n.toString(2).padStart(10, "0")
105 .replace(/^([0-9]{5})/, '$1x')
106 .replace(/0/g, '·')
107 .replace(/1/g, '+');
108 return changeCharColor(str, heat, word);
109 }
110
111 % use Mojo::ByteStream 'b';
112 var paraResults = <%= b(Mojo::JSON::to_json($lists)) %>;
113 var urlprefix = new URLSearchParams(window.location.search);
114 if (paraResults.length > 0 && paraResults[0] != null) {
115 var nvecs = [],
116 nwords = [],
117 nranks = [],
118 nmarked = [];
119 for(var i = 0; i < paraResults.length; i++) {
120 nwords = nwords.concat(paraResults[i].map(function(a){return a.word;}));
121 nvecs = nvecs.concat(paraResults[i].map(function(a){return a.vector;}));
122 nranks = nranks.concat(paraResults[i].map(function(a){return a.rank;}));
123 nmarked = nmarked.concat(paraResults[i].map(function(a){return a.marked;}));
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100[diff] [blame]124 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]125 showMap({target: " "+urlParams.get('word')+" ", mergedEnd: 0, words: nwords, vecs: nvecs, ranks: nranks, marked: nmarked} );
126 var t = $('#firsttable').DataTable({
127 data: paraResults[0],
128 "sScrollY": "780px",
129 "bScrollCollapse": true,
130 "bPaginate": false,
131 "bJQueryUI": true,
132 "dom": '<"top">rt<"bottom"flp><"clear">',
133 "columns": [
134 { "data": "rank", type: "allnumeric" },
135 { "data": "dist", render: function ( data, type, row ) {return data.toFixed(3) }},
136 { "data": "word", render: function ( data, type, row ) {urlprefix.set("word", data); return '<a href="?' + urlprefix + '">' + data + '</a>' }}
137 ],
138 "columnDefs": [
139 { className: "dt-right", "targets": [0,1] },
140 { "searchable": false,
141 "orderable": false,
142 "targets": 0
143 },
144 { "orderSequence": [ "desc" ], "targets": [ 1 ] },
145 { "orderSequence": [ "asc", "desc" ], "targets": [ 2 ] },
146 ],
147 "oLanguage": {
148 "sSearch": "Filter: "
149 },
150 "order": [[ 1, 'desc' ]],
151 } );
Marc Kupietz0d4c0ca2017-12-04 09:18:56 +0100[diff] [blame]152
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]153 t.on( 'order.dt search.dt', function () {
154 t.column(0, {order:'applied'}).nodes().each( function (cell, i) {
155 cell.innerHTML = i+1;
156 } );
157 } ).draw();
158
159 $( "#first" ).clone().prependTo( "#tabs-2" );
160
161 }
162
163
164 var collocatorData = <%= b(Mojo::JSON::to_json($collocators)) %>;
165 var maxHeat; // = Math.max.apply(Math,collocatorData.map(function(o){return o.cprob;}))
166
167 if (collocatorData != null) {
168 maxHeat = Math.max.apply(Math,collocatorData.map(function(o){return Math.max.apply(Math,o.heat);}))
169 collocatorTable = $('#secondtable').DataTable({
170 data: collocatorData,
171 "sScrollY": "780px",
172 "bScrollCollapse": true,
173 "bPaginate": false,
174 "bJQueryUI": true,
175 "dom": '<"top">rt<"bottom"flp><"clear">',
176 "columns": [
177 { "data": "rank", type: "allnumeric" },
178 { "data": "pos", width: "7%", sClass: "dt-center mono compact", render: function ( data, type, row ) {return bitvec2window(data, row.heat, row.word) }},
179 { "data": "max", render: function ( data, type, row ) {return data.toFixed(3) }},
180 { "data": "average", render: function ( data, type, row ) {return data.toFixed(3) }},
181 { "data": "prob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
182 { "data": "cprob", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
183 { "data": "overall", type: "scientific", render: function ( data, type, row ) {return data.toExponential(3) } },
184 { "data": "word", sClass: "collocator" },
185 { "data": "rank", type: "allnumeric" }
186 ],
187 "columnDefs": [
188 { className: "dt-right", "targets": [0,2,3,4,5,6] },
189 { className: "dt-center", "targets": [ 1] },
190 { "searchable": false,
191 "orderable": false,
192 "targets": [0, 8]
193 },
194 { "type": "scientific", targets: [2,3,4,5,6] },
195 { "orderSequence": [ "desc" ], "targets": [ 2, 3, 4, 5, 6 ] },
196 { "orderSequence": [ "asc", "desc" ], "targets": [ 1, 7 ] },
197 { "targets": [8], "visible": false }
198 ],
199 "oLanguage": {
200 "sSearch": "Filter: "
201 },
202 "order": [[ 4, 'desc' ]],
203 } );
204 $.ajaxSetup({
205 type: 'POST',
206 timeout: 30000,
207 error: function(xhr) {
208 $('#display_error')
209 .html('Error: ' + xhr.status + ' ' + xhr.statusText);
210 }
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]211 });
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]212
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]213 classicCollocatorTable = $('#classicoloctable').DataTable({
214 ajax: {
Marc Kupietz8893dec2018-01-24 09:52:02 +0100[diff] [blame]215 method: "GET",
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]216 url: '/derekovecs/getClassicCollocators',
217 dataType: 'json',
218 dataSrc: "",
219 timeout: 30000,
Marc Kupietz8893dec2018-01-24 09:52:02 +0100[diff] [blame]220 data: { w: paraResults[0][0].rank }
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]221 },
Marc Kupietzad783722018-01-13 17:45:21 +0100[diff] [blame]222 "initComplete":function(settings, json){
223 $("td.collocator").click(function(){
224 queryKorAPCII(this.textContent + " /w5 " + CIIsearchWords);
225 });
226 },
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]227 "sScrollY": "780px",
228 "bScrollCollapse": true,
229 "bPaginate": false,
230 "bJQueryUI": true,
231 "dom": '<"top">rt<"bottom"flp><"clear">',
232 "columns": [
233 // { "data": "pos", width: "7%", sClass: "dt-center mono compact", render: function ( data, type, row ) {return bitvec2window(data, row.heat, row.word) }},
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]234 { "data": "llr", render: function ( data, type, row ) {return data.toFixed(0) }},
235 { "data": "lfmd", render: function ( data, type, row ) {return data.toFixed(1) }},
Marc Kupietzcfcdcfc2018-01-24 09:51:26 +0100[diff] [blame]236// { "data": "fpmi", type: "scientific", render: function ( data, type, row ) {return data.toExponential(2) } },
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]237 { "data": "npmi", render: function ( data, type, row ) {return data.toFixed(2) }},
238 { "data": "llfmd", render: function ( data, type, row ) {return data.toFixed(1) }},
239 { "data": "rlfmd", render: function ( data, type, row ) {return data.toFixed(1) }},
240 { "data": "lnpmi", render: function ( data, type, row ) {return data.toFixed(2) }},
241 { "data": "rnpmi", render: function ( data, type, row ) {return data.toFixed(2) }},
242 { "data": "f" },
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]243 { "data": "word", sClass: "collocator" }
244 ],
245 "columnDefs": [
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]246 { className: "dt-right", "targets": [0,1,2,3,4,5,6,7] },
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]247 { "searchable": false,
248 "orderable": false,
249 "targets": []
250 },
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]251 { "orderSequence": [ "desc" ], "targets": [ 0, 1, 2,3,4,5,6,7 ] },
252 { "orderSequence": [ "asc", "desc" ], "targets": [ 8 ] },
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]253 ],
254 "oLanguage": {
255 "sSearch": "Filter: "
256 },
257 "order": [[ 1, 'desc' ]],
258 });
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]259
Marc Kupietzad783722018-01-13 17:45:21 +0100[diff] [blame]260 $("td.collocator").click(function(){
261 queryKorAPCII(this.textContent + " /w5 " + CIIsearchWords);
262 });
263
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]264 collocatorTable.on( 'order.dt search.dt', function () {
265 collocatorTable.column(0, {order:'applied'}).nodes().each( function (cell, i) {
266 cell.innerHTML = i+1;
267 } );
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]268 }).draw();
269 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]270
Marc Kupietz78b434a2018-01-12 22:33:32 +0100[diff] [blame]271 if (localStorage && !window.location.hash) { // let's not crash if some user has IE7
272 var index = parseInt(localStorage['tab']||'0');
273 $("#tabs").tabs({ active: index });
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]274 }
275 $("#tabs").css("visibility", "visible"); // now we can show the tabs
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]276 });
277
278 $(function(){
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]279 $("#dropdownoptions").dialog({
280 title: "Options",
281 autoOpen: false,
282 modal: false,
283 draggable: false,
284 height: "auto",
285 width: "auto",
286 resizable: false,
287 buttons: {
288 "Cancel": function() {
289 $( this ).dialog( "close" );
290 },
291 "Apply": function() {
292 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
293 }
294 }
295 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]296 });
297
298 $(function(){
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]299 $("#showoptions").click(function(){
300 $("#dropdownoptions").dialog("open");
301 var target = $(this);
302 $("#dropdownoptions").dialog("widget").position({
303 my: 'left bottom',
304 at: 'left bottom',
305 of: target
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]306 });
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]307 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]308 });
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]309
310 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]311 $( "#no_iterations" ).spinner({
312 spin: function( event, ui ) {
313 if ( ui.value < 1000 ) {
314 $( this ).spinner( "value", 1000 );
315 return false;
316 } else if ( ui.value > 10000 ) {
317 $( this ).spinner( "value", 10000 );
318 return false;
319 }
320 }
321 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]322 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]323
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]324 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]325 $( "#neighbours" ).spinner({
326 spin: function( event, ui ) {
327 if ( ui.value < 0 ) {
328 $( this ).spinner( "value", 0 );
329 return false;
330 } else if ( ui.value > 200 ) {
331 $( this ).spinner( "value", 200 );
332 return false;
333 }
334 }
335 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]336 } );
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]337
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]338 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]339 $( "#cutoff" ).spinner({
340 spin: function( event, ui ) {
341 if ( ui.value < 100000 ) {
342 $( this ).spinner( "value", 100000 );
343 return false;
344 } else if ( ui.value > 2000000 ) {
345 $( this ).spinner( "value", 2000000 );
346 return false;
347 }
348 }
349 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]350 } );
351
352 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]353 $( "#tabs" ).tabs().addClass('tabs-min');
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]354 } );
355
356 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]357 $( ".controlgroup-vertical" ).controlgroup({
358 "direction": "vertical"
359 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]360 } );
361
362 $(function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]363 $( document ).tooltip({
364 content: function() {
365 return $(this).attr('title');
366 }}
367 )
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]368 });
Marc Kupietz694610d2017-11-25 18:30:03 +0100[diff] [blame]369
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]370 $(function () {
371 $(document).tooltip({
372 content: function () {
373 return $(this).prop('title');
374 },
375 show: null,
376 close: function (event, ui) {
377 ui.tooltip.hover(
378 function () {
379 $(this).stop(true).fadeTo(400, 1);
380 },
381 function () {
382 $(this).fadeOut("400", function () {
383 $(this).remove();
384 })
385 });
386 }
387 });
388 });
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]389 </script>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]390 <script src="//d3js.org/d3.v3.min.js" charset="utf-8"></script>
391 <script src="/derekovecs/js/tsne.js"></script>
392 <script src="/derekovecs/js/som.js"></script>
393 <script src="/derekovecs/js/labeler.js"></script>
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]394 <style>
395 body, input {
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]396 font-family: Lato, sans-serif;
397 font-size: 11pt;
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]398 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100[diff] [blame]399
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]400 .info {
401 font-size: 8pt;
402 margin-top: 4px;
403 /* position: absolute;
404 bottom: 0;
405 left: 0;
406 right: 0; */
407 }
408
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]409 h1, h2, h3 {
410 margin: 5px 10px 0 0;
411 color: rgb(246,168,0);
412 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;
413 font-weight: bold;
414 line-height: 1.35;
415 letter-spacing: normal;
416 text-transform: uppercase;
417 text-shadow: none;
418 word-wrap: break-word;
419 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]420
421
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]422 showoptions, #SEARCH {
423 margin-left: 10px;
424 margin-right: 10px;
425 }
Marc Kupietz6dbadd12017-11-29 16:43:33 +0100[diff] [blame]426
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]427 .tabs-left-vertical .ui-tabs-nav {
428 position: absolute;
429 width: 21em;
430 transform: translate(-100%,0%) rotate(-90deg);
431 transform-origin: 100% 0%;
432 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100[diff] [blame]433
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]434 .tabs-left-vertical .ui-tabs-nav li {
435 float: right;
436 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100[diff] [blame]437
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]438 .tabs-left-vertical .ui-tabs-panel {
439 padding-left: 3.5em;
440 }
Marc Kupietzdab9f222017-11-29 14:22:59 +0100[diff] [blame]441
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]442 .tabs-left-vertical .ui-tabs-panel {
443 height: 20em;
444 }
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100[diff] [blame]445
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]446 .mono {
Marc Kupietzc8221182017-12-08 17:26:19 +0100[diff] [blame]447 font-family: "DejaVu Sans Mono", Inconsolata, SourceCodePro, "Courier New", Courier, monospace;
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]448 }
Marc Kupietz30ca4342017-11-22 21:21:20 +0100[diff] [blame]449
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]450 .ui-tooltip-content {
Marc Kupietz4116b432017-12-06 14:15:32 +0100[diff] [blame]451 font-size: 10pt;
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]452 color: #222222;
453 }
Marc Kupietzf4b49392016-04-28 10:49:56 +0200[diff] [blame]454
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]455 svg > .ui-tooltip-content {
456 font-size: 8pt;
457 color: #222222;
458 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]459
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]460 a.merged {
461 color: green;
462 fill: green;
463 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]464
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]465 #first a {
466 text-decoration: none;
467 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]468
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]469 a.marked, #first a.marked {
470 text-decoration: underline;
471 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]472
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]473 a.target {
474 color: red;
475 fill: red;
476 }
Marc Kupietz694610d2017-11-25 18:30:03 +0100[diff] [blame]477
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]478 table.display {
479 width: 40% important!;
480 margin: 1; /* <- works for me this way ****/
481 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]482
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]483 table.dataTable thead th, table.dataTable thead td, table.dataTable tbody td {
484 padding: 2px 2px;
485 // border-bottom: 1px solid #111;
486 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]487
Marc Kupietz8f75b1f2017-12-06 09:42:13 +0100[diff] [blame]488 td.collocator {
489 cursor: pointer;
490 }
491
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]492 #collocators {
493 margin-bottom: 15px;
494 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]495
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]496 #header {
497 width: 100%;
498 // border: 1px solid red;
499 overflow: hidden; /* will contain if #first is longer than #second */
500 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100[diff] [blame]501
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]502 #topwrapper {
503 width: 100%;
504 // border: 1px solid red;
505 overflow: hidden; /* will contain if #first is longer than #second */
506 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]507
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]508 #wrapper {
509 // border: 1px solid red;
510 overflow: hidden; /* will contain if #first is longer than #second */
511 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]512
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]513 #pagetitle {
514 max-width: 460px;
515 margin-right: 20px;
516 float: left;
517 overflow: hidden; /* if you don't want #second to wrap below #first */
518 // border: 1px solid green;
519 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]520
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]521 #options {
522 float: left;
523 width: 800px;
524 margin: 10px;
525 overflow: hidden; /* if you don't want #second to wrap below #first */
526 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100[diff] [blame]527
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]528 #word {
529 width: 50%;
530 }
Marc Kupietz81bf39c2017-11-29 17:04:35 +0100[diff] [blame]531
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]532 #first {
533 margin-right: 20px;
534 float: left;
535 overflow: hidden; /* if you don't want #second to wrap below #first */
536 // border: 1px solid green;
537 }
538 #tabs {
539 margin-right: 20px;
540 overflow: hidden; /* if you don't want #second to wrap below #first */
541 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]542
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]543 .tabs-min {
544 background: transparent;
545 border: none;
546 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]547
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]548 .tabs-min .ui-widget-header {
549 background: transparent;
550 border: none;
551 border-bottom: 1px solid #c0c0c0;
552 -moz-border-radius: 0px;
553 -webkit-border-radius: 0px;
554 border-radius: 0px;
555 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]556
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]557 .tabs-min .ui-tabs-nav .ui-state-default {
558 background: transparent;
559 border: none;
560 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]561
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]562 .tabs-min .ui-tabs-nav .ui-state-active {
563 background: transparent url(derekovecs/img/uiTabsArrow.png) no-repeat bottom center;
564 border: none;
565 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]566
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]567 .tabs-min .ui-tabs-nav .ui-state-default a {
568 color: #c0c0c0;
569 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]570
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]571 .tabs-min .ui-tabs-nav .ui-state-active a {
572 color: rgb(246,168,0);
573 }
Marc Kupietzdf3d4b52017-11-29 16:57:27 +0100[diff] [blame]574
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]575 #embed {
576 max-width: 802px;
577 border: 1px solid #333;
578 }
Marc Kupietz4abcd682017-11-28 20:51:08 +0100[diff] [blame]579
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]580 #second {
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]581 min-width: 700px;
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]582 // border: 1px solid #333;
583 overflow: hidden; /* if you don't want #second to wrap below #first */
584 }
585 #som2 svg {
586 border: 1px solid #333;
587 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]588
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]589 #cost {
590 font-size: 8pt;
591 color: #222222;
592 margin-top: 4px;
593 margin-bottom: 12px;
594 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]595
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]596 #sominfo1, #sominfo {
597 font-size: 8pt;
598 color: #222222;
599 margin-top: 0px;
600 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]601
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]602 #somcolor1, #somcolor2, #somcolor3 {
603 display: inline-block;
604 height: 10px;
605 width: 10px;
606 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]607
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]608 #third {
609 border: 1px solid #333;
610 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]611
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]612 th {
613 color: rgb(246,168,0);
614 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;
615 }
616
617 p.citation {
618 font-size: 8pt;
619 color: #999999;
620 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]621 </style>
622 <script>
623
624 var opt = {epsilon: <%= $epsilon %>, perplexity: <%= $perplexity %>},
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]625 mapWidth = 800, // width map
626 mapHeight = 800,
627 jitterRadius = 7;
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]628
629 var T = new tsnejs.tSNE(opt); // create a tSNE instance
630
631 var Y;
632
633 var data;
634 var labeler;
635
636 function applyJitter() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]637 svg.selectAll('.tsnet')
638 .data(labels)
639 .transition()
640 .duration(50)
641 .attr("transform", function(d, i) {
642 T.Y[i][0] = (d.x - mapWidth/2 - tx)/ss/20;
643 T.Y[i][1] = (d.y - mapHeight/2 - ty)/ss/20;
644 return "translate(" +
645 (d.x) + "," +
646 (d.y) + ")";
647 });
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]648 }
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]649
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]650 function updateEmbedding() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]651 var Y = T.getSolution();
652 svg.selectAll('.tsnet')
653 .data(data.words)
654 .attr("transform", function(d, i) {
655 return "translate(" +
656 ((Y[i][0]*20*ss + tx) + mapWidth/2) + "," +
657 ((Y[i][1]*20*ss + ty) + mapHeight/2) + ")"; });
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]658 }
659
660 var svg;
661 var labels = [];
662 var anchor_array = [];
663 var text;
664
665 function drawEmbedding() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]666 var urlprefix = new URLSearchParams(window.location.search);
667 urlprefix.delete("word");
668 urlprefix.append("word","");
Marc Kupietz8f9c86a2017-12-04 17:17:13 +0100[diff] [blame]669
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]670 $("#embed").empty();
671 var div = d3.select("#embed");
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]672
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]673 // get min and max in each column of Y
674 var Y = T.Y;
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]675
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]676 svg = div.append("svg") // svg is global
677 .attr("width", mapWidth)
678 .attr("height", mapHeight);
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]679
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]680 var g = svg.selectAll(".b")
681 .data(data.words)
682 .enter().append("g")
683 .attr("class", "tsnet");
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]684
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]685 g.append("a")
686 .attr("xlink:href", function(word) {
687 return "?"+urlprefix+word; })
688 .attr("class", function(d, i) {
689 var res="";
690 if(data.marked[i]) {
691 res="marked ";
692 }
693 if(data.target.indexOf(" "+d+" ") >= 0) {
694 return res+"target";
695 } else if(data.ranks[i] < data.mergedEnd) {
696 return res+"merged";
697 } else {
698 return res;
699 }
700 })
701 .attr("title", function(d, i) {
702 if(data.mergedEnd > 0) {
703 if(data.ranks[i] >= data.mergedEnd) {
704 return "rank: "+i +" "+"freq. rank: "+(data.ranks[i]).toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
705 } else {
706 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",") + " (merged vocab)";
707 }
708 } else {
709 return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",");
710 }
711 })
712 .append("text")
713 .attr("text-anchor", "top")
714 .attr("font-size", 12)
715 .text(function(d) { return d; });
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]716
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]717 var zoomListener = d3.behavior.zoom()
718 .scaleExtent([0.1, 10])
719 .center([0,0])
720 .on("zoom", zoomHandler);
721 zoomListener(svg);
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]722 }
723
724 var tx=0, ty=0;
725 var ss=1;
726 var iter_id=-1;
727
728 function zoomHandler() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]729 tx = d3.event.translate[0];
730 ty = d3.event.translate[1];
731 ss = d3.event.scale;
732 updateEmbedding();
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]733 }
734
735 var stepnum = 0;
736
737 function stopStep() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]738 clearInterval(iter_id);
739 text = svg.selectAll("text");
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]740
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]741 // jitter function needs different data and co-ordinate representation
742 labels = d3.range(data.words.length).map(function(i) {
743 var x = (T.Y[i][0]*20*ss + tx) + mapWidth/2;
744 var y = (T.Y[i][1]*20*ss + ty) + mapHeight/2;
745 anchor_array.push({x: x, y: y, r: jitterRadius});
746 return {
747 x: x,
748 y: y,
749 name: data.words[i]
750 };
751 });
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]752
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]753 // get the actual label bounding boxes for the jitter function
754 var index = 0;
755 text.each(function() {
756 labels[index].width = this.getBBox().width;
757 labels[index].height = this.getBBox().height;
758 index += 1;
759 });
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]760
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]761
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]762 // setTimeout(updateEmbedding, 1);
763 // setTimeout(
764 labeler = d3.labeler()
765 .label(labels)
766 .anchor(anchor_array)
767 .width(mapWidth)
768 .height(mapHeight)
769 .update(applyJitter);
770 // .start(1000);
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]771
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]772 iter_id = setInterval(jitterStep, 1);
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]773 }
774
775 var jitter_i=0;
776
777 function jitterStep() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]778 if(jitter_i++ > 100) {
779 clearInterval(iter_id);
780 } else {
781 labeler.start2(10);
782 applyJitter();
783 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]784 }
785
786 var last_cost=1000;
787
788 function step() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]789 var i = T.iter;
Marc Kupietz34c08172017-11-29 17:08:47 +0100[diff] [blame]790
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]791 if(i > <%= $no_iterations %>) {
792 stopStep();
793 } else {
794 var cost = Math.round(T.step() * 100000) / 100000; // do a few steps
795 $("#cost").html("tsne iteration " + i + ", cost: " + cost.toFixed(5));
796 if(i % 250 == 0 && cost >= last_cost) {
797 stopStep();
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]798 } else {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]799 last_cost = cost;
800 updateEmbedding();
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]801 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]802 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]803 }
804
805 function showMap(j) {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]806 data=j;
807 T.iter=0;
808 iter_id = -1;
809 last_cost=1000;
810 T.initDataRaw(data.vecs); // init embedding
811 drawEmbedding(); // draw initial embedding
Marc Kupietz78114532017-11-29 17:00:16 +0100[diff] [blame]812
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]813 if(iter_id >= 0) {
814 clearInterval(iter_id);
815 }
816 //T.debugGrad();
817 iter_id = setInterval(step, 1);
818 if(true) { // (<%= $show_som %>) {
819 makeSOM(j, <%= $no_iterations %>);
820 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]821 }
Marc Kupietz39179ab2017-07-04 16:28:06 +0200[diff] [blame]822 var queryword;
823
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]824 function showCollocatorSOM() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]825 if (collocatorTable) {
826 var ctableData = collocatorTable.rows().data();
827 var nwords = [],
828 nranks = [];
829 for (var i=0; i < ctableData.length && i < 100; i++) {
830 nranks.push(ctableData[i].rank);
831 nwords.push(ctableData[i].word);
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]832 }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]833 $.post('/derekovecs/getVecsByRanks',
834 JSON.stringify(nranks),
835 function(data, status){
836 showMap({target: " "+urlParams.get('word')+" ", mergedEnd: 0, words: nwords, vecs: data, ranks: nranks, marked: Array(100).fill(false)} );
837 }, 'json');
838 }
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]839 }
840
Marc Kupietz39179ab2017-07-04 16:28:06 +0200[diff] [blame]841 function onload() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]842 queryword = document.getElementById('word');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200[diff] [blame]843 }
844
845 function queryKorAP() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]846 window.open('http://korap.ids-mannheim.de/kalamar/?q='+queryword.value, 'KorAP');
Marc Kupietz39179ab2017-07-04 16:28:06 +0200[diff] [blame]847 }
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100[diff] [blame]848
849 function queryKorAPCII(query) {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]850 window.open('http://korap.ids-mannheim.de/kalamar/?ql=cosmas2&q='+query, 'KorAP');
Marc Kupietz4dc270c2017-11-24 10:17:12 +0100[diff] [blame]851 }
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]852 </script>
853 </head>
Marc Kupietz39179ab2017-07-04 16:28:06 +0200[diff] [blame]854 <body onload="onload()">
Marc Kupietz5a7f9ac2018-01-30 11:22:44 +0100[diff] [blame]855 <div style="display:none;" id="lfmd_ttt">PMI cubed [1], also called log-frequency biased mutual dependency [2]: $$\text{PMI}^3=\text{LFMD}=log_2\frac{p^3(w_1,w_2)}{p(w_1) p(w_2)}$$<p class="citation">[1] Daille, B. (1994): <a href="http://www.bdaille.com/index.php?option=com_docman&task=doc_download&gid=8&Itemid=">Approche mixte pour l’extraction automatique de terminologie: statistiques lexicales et filtres linguistiques</a>. PhD thesis, Université Paris 7.</p><p class="citation">[2] Thanopoulos, A., Fakotakis, N., Kokkinakis, G. (2002): <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.11.8101&rep=rep1&type=pdf">Comparative evaluation of collocation extraction metrics</a>. In: Proc. of LREC 2002: 620–625.</p></div>
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]856 <div style="display:none;" id="npmi_ttt">Normalized pointwise mutual information: $$\frac{log_2\frac{p(w_1,w_2)}{p(w_1)p(w_2)}}{-log_2(p(w_1,w_2))}$$<p class="citation">Bouma, Gerlof (2009): <a href="https://svn.spraakdata.gu.se/repos/gerlof/pub/www/Docs/npmi-pfd.pdf">Normalized (pointwise) mutual information in collocation extraction</a>. In Proceedings of GSCL.</p></div>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]857 <div id="header">
858 <div id="pagetitle">
859 <h1>DeReKo-Vectors</h1>
Marc Kupietz4fcda0c2017-11-29 09:00:31 +0100[diff] [blame]860 </div>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]861 <div id="options" class="widget">
862 <form id="queryform">
863 <input id="word" type="text" name="word" placeholder="Word(s) to be searched" value="<%= $word %>"
864 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."/>
865 <input id="SEARCH" type="button" value="SEARCH">
866 <input type="button" id="showoptions" name="showoptions" value="Options" />
867 </form>
868 <div id="dropdownoptions" style="display: none">
869 <form id="optionsform">
870 <div class="controlgroup-vertical">
871 <label for="cutoff">cut-off</label>
872 <input id="cutoff" type="text" name="cutoff" size="10" value="<%= $cutoff %>" title="Only consider the most frequent x word forms.">
873 <label for="dedupe">dedupe</label>
874 <input id="dedupe" type="checkbox" name="dedupe" value="1" <%= ($dedupe ? "checked" : "") %> title="radically filter out any near-duplicates">
875 % if($mergedEnd > 0) {
876 <label for="sbf">backw.</label>
877 <input id="sbf" type="checkbox" name="sbf" value="1" <%= ($searchBaseVocabFirst ? "checked" : "") %> title="If checkecked base vocabulary will be searched first. Otherwise merged vocabulray will be searched first.">
878 % }
879 <label for="neighbours">max. neighbours:</label>
880 <input id="neighbours" size="4" name="n" value="<%= $no_nbs %>">
881 <label for="no_iterations">max. iterations</label>
882 <input id="no_iterations" name="N" size="4" value="<%= $no_iterations %>">
883 <!-- <label for="dosom">SOM</label>
884 <input id="dosom" type="checkbox" name="som" value="1" <%= ($show_som ? "checked" : "") %>> -->
885 % if($collocators) {
886 <label for="sortby">window/sort</label>
887 <select id="sortby" name="sort">
888 <option value="0" <%= ($sort!=1 && $sort!=2? "selected":"") %>>auto focus</option>
889 <!-- <option value="1" <%= ($sort==1? "selected":"") %>>any single position</option>
890 <option value="2" <%= ($sort==2? "selected":"") %>>whole window</option> -->
891 </select>
892 % }
893 <input type="button" value="→ KorAP" onclick="queryKorAP();" title="query word with KorAP"/>
894 </div>
895 </form>
896 </div>
897 </div>
898 </div>
899 <div id="topwrapper">
900 <div style="visibility: hidden;" id="tabs">
901 <ul>
902 <li><a href="#tabs-1">Semantics (TSNE-map)</a></li>
903 <li><a href="#tabs-2">Semantics (SOM)</a></li>
904 <li><a href="#tabs-3">Syntagmatic (collocators)</a></li>
905 </ul>
906 <div id="tabs-1">
907 % if($lists && (@$lists) > 0 && (@$lists)[0]) {
908 <div id="wrapper">
909 <div id="first" style="width: 320px">
910 <table class="display compact nowrap" id="firsttable">
911 <thead>
912 <tr>
913 <th align="right">#</th><th align="right">cos</th><th align="left">similars</th>
914 </tr>
915 </thead>
916 <tbody>
917 <tr>
918 <td align="right">
919 </td>
920 <td align="right">
921 </td>
922 <td></td>
923 </tr>
924 </tbody>
925 </table>
926 </div>
927 <script>
928 </script>
929 <div id="second">
930 <div id="embed">
931 </div>
932 <div id="cost">
933 </div>
934 </div>
935 </div>
936 % } elsif($word !~ /^\s*$/) {
937 <div id="wrapper">
938 <div id="not-found-dialog" title="Not found">
939 <p>ERROR: "<%= $word %>" not found in vocabluary.</p>
940 <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>
941 </div>
942 <script>
943 $( function() {
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]944 $( "#not-found-dialog" ).dialog({
945 autoOpen: true,
946 modal: true,
947 draggable: false,
948 height: "auto",
949 width: "auto",
950 resizable: false,
951 buttons: {
952 "OK": function() {
953 $( this ).dialog( "close" );
954 },
955 "Apply": function() {
956 window.open($(location).attr('pathname')+'?'+$('form').serialize(), "_self");
957 }
958 }
959 });
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]960 });
961 </script>
962 </div>
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]963 % }
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]964 </div>
965 <div id="tabs-2">
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]966 <div id="som2" style="width: 800;">
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]967 </div>
968 <div id="sominfo1"><span id="somcolor1"> </span> <span id="somword1"> </span> <span id="somcolor2"> </span> <span id="somword2"> </span> <span id="somcolor3"> </span></div>
969 <div id="sominfo">SOM iteration <span id="iterations">0</span></div>
970 </div>
Marc Kupietzcfcdcfc2018-01-24 09:51:26 +0100[diff] [blame]971 <div id="tabs-3" style="width:1500px">
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]972 <div style="width: 45%; float: left;" id="secondt">
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]973 <table class="display compact nowrap" id="secondtable">
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]974 <thead>
975 <tr>
976 % if($collocators) {
977 <th>#</th>
Marc Kupietz4116b432017-12-06 14:15:32 +0100[diff] [blame]978 <th align="center" title="Activation of the respective collocator in the columns around the target normalized by its maximum (red). Columns selected by the auto-focus funtion (which window of all possible column-combinations maximizes ⊥(a/c)?) are marked with +. Click on the column postions to lauch a KorAP query with target word and collocator in the respective position.">w'</th>
979 <th align="right" title="Maximum activation of the collocator anywhere in the output layer.">max(a)</th>
980 <th title="Average raw activation of the collocator in the columns selected by auto-focus." align="right">⟨a⟩</th>
981 <th title="Sum of activations over the selected colunns normalized by the total activation sum of the selected columns." align="right">Σa/Σw'</th>
982 <th title="Co-norm of the column-normalized activations over the colunns selected by the auto-focus." align="right">⊥(a/c)</th>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]983 <th title="Sum of the activations over the whole window normalized by the total window sum (no auto-focus)." align="right">Σa/Σw</th>
984 <th align="left">collocator</th>
985 % }
986 </tr>
987 </thead>
988 <tbody>
989 <tr>
990 <td align="right">
991 </td>
992 <td align="right">
993 </td>
994 <td align="right">
995 </td>
996 <td align="right">
997 </td>
998 <td align="right">
999 </td>
1000 <td align="right">
1001 </td>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]1002 </tr>
1003 </tbody>
1004 </table>
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]1005 </div>
Marc Kupietza0ffb392018-01-25 08:53:43 +0100[diff] [blame]1006 <div id="ccd" style="margin-left:0px; float:right; width:800px;">
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]1007 <table class="display compact nowrap" id="classicoloctable">
1008 <thead>
Marc Kupietzcd136e22018-01-29 09:18:12 +0100[diff] [blame]1009 % if($collocators) {
1010 <tr>
1011 <th>LLR</th>
Marc Kupietz5a7f9ac2018-01-30 11:22:44 +0100[diff] [blame]1012 <th id="lfmd_tt">PMI³</th>
Marc Kupietzcd136e22018-01-29 09:18:12 +0100[diff] [blame]1013 <th id="npmi_tt">nPMI</th>
Marc Kupietz5a7f9ac2018-01-30 11:22:44 +0100[diff] [blame]1014 <th title="PMI³ restricted to left neighbour">l-PMI³</th>
1015 <th title="PMI³ restricted to right neighbour">r-PMI³</th>
1016 <th title="nPMI restricted to left neighbour">l-nPMI</th>
1017 <th title="nPMI restricted to right neighbour">r-nPMI</th>
Marc Kupietzcd136e22018-01-29 09:18:12 +0100[diff] [blame]1018 <th title="raw max frequency of collocation within window">raw</th>
1019 <th>collocator</th>
1020 </tr>
1021 % }
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]1022 </thead>
1023 <tbody>
1024 <tr>
1025 <td align="right">
1026 </td>
1027 <td align="right">
1028 </td>
1029 <td align="right">
1030 </td>
1031 <td align="right">
1032 </td>
Marc Kupietzcfcdcfc2018-01-24 09:51:26 +0100[diff] [blame]1033 <td align="right">
1034 </td>
1035 <td align="right">
1036 </td>
Marc Kupietzb4b53ca2018-01-25 08:57:11 +0100[diff] [blame]1037 <td align="right"></td>
1038 <td align="right"></td>
1039 <td align="left"></td>
Marc Kupietze6a7a732018-01-12 09:21:08 +0100[diff] [blame]1040 </tr>
1041 </tbody>
1042 </table>
1043 </div>
1044 <div style="clear:both" ></div>
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]1045 <div style="float: right; overflow: hidden" id="extra"><button onClick="showCollocatorSOM()"> </button></div>
Marc Kupietz66bfd952017-12-11 09:59:45 +0100[diff] [blame]1046 </div>
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]1047 </div>
1048 </div> <!-- topwrapper -->
1049 <div style="clear: both;"></div>
1050 </div>
Marc Kupietzebea4702018-01-12 09:16:09 +0100[diff] [blame]1051 % if($training_args && (@$lists)[0]) {
1052 <div class="info">
1053 % if($training_args =~ /-type\s*3/) {
1054 Calculations are based on a word embedding model trained with an extension of <a href="https://github.com/wlin12/wang2vec/">wang2vec</a> using the following parameters: <%= $training_args %>
1055 % } else {
1056 Calculations are based on a word embedding model trained with <a href="https://code.google.com/p/word2vec/">word2vec</a> using the following parameters: <%= $training_args %>
1057 % }
1058 </div>
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]1059 % }
Marc Kupietz58270662017-12-04 12:10:06 +0100[diff] [blame]1060 </body>
Marc Kupietz83305222016-04-28 09:57:22 +0200[diff] [blame]1061</html>