DMVCCM.html

   1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
   2                "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml"
   4 lang="en" xml:lang="en">
   5 <head>
   6 <title>DMV/CCM &ndash; todo-list / progress</title>
   7 <meta http-equiv="Content-Type" content="text/html;charset=utf-8"/>
   8 <meta name="generator" content="Org-mode"/>
   9 <meta name="generated" content="2008/06/06 19:12:56"/>
  10 <meta name="author" content="Kevin Brubeck Unhammer"/>
  11 <link rel="stylesheet" type="text/css" href="http://www.student.uib.no/~kun041/org.css">
  12 </head><body>
  13 <h1 class="title">DMV/CCM &ndash; todo-list / progress</h1>
  14
  15
  16 <div id="table-of-contents">
  17 <h2>Table of Contents</h2>
  18 <div id="text-table-of-contents">
  19 <ul>
  20 <li><a href="#sec-1">1 dmvccm report and project</a></li>
  21 <li><a href="#sec-2">2 Adjacency and combining it with inner()</a></li>
  22 <li><a href="#sec-3">3 [#A] P_STOP for IO/EM</a></li>
  23 <li><a href="#sec-4">4 P_CHOOSE for IO/EM</a></li>
  24 <li><a href="#sec-5">5 Initialization   </a></li>
  25 <li><a href="#sec-6">6 [#C] Deferred</a></li>
  26 <li><a href="#sec-7">7 Expectation Maximation in IO/DMV-terms</a></li>
  27 <li><a href="#sec-8">8 Python-stuff</a></li>
  28 <li><a href="#sec-9">9 Git</a></li>
  29 </ul>
  30 </div>
  31 </div>
  32
  33 <div id="outline-container-1" class="outline-2">
  34 <h2 id="sec-1">1 dmvccm report and project</h2>
  35 <div id="text-1">
  36
  37 <p><span class="timestamp-kwd">DEADLINE: </span> <span class="timestamp">2008-06-30 Mon</span><br/>
  38 But absolute, extended, really-quite-dead-now deadline: August
  39 31&hellip;
  40 </p><ul>
  41 <li>
  42 <a href="src/dmv.py">dmv.py</a>
  43 </li>
  44 <li>
  45 <a href="src/io.py">io.py</a>
  46 </li>
  47 <li>
  48 <a href="src/harmonic.py">harmonic.py</a>
  49 </li>
  50 </ul>
  51 </div>
  52
  53 </div>
  54
  55 <div id="outline-container-2" class="outline-2">
  56 <h2 id="sec-2">2 <span class="todo">TODO</span> Adjacency and combining it with inner()</h2>
  57 <div id="text-2">
  58
  59 <p>Each DMV_Rule now has both a probN and a probA, for
  60 adjacencies. inner() needs the correct one in each case.
  61 </p>
  62 <p>
  63 Adjacency gives a problem with duplicate words/tags, eg. in the
  64 sentence "a a b". If this has the dependency structure b-&gt;a<sub>0</sub>-&gt;a<sub>1</sub>,
  65 then b is non-adjacent to a<sub>0</sub> and should use probN (for the LRStop and
  66 the attachment of a<sub>0</sub>), while the other rules should all use
  67 probA. But within the e(0,2,b) we can't just say "oh, a has index 0
  68 so it's not adjacent to 2", since there's also an a at index 1, and
  69 there's also a dependency structure b-&gt;a<sub>1</sub>-&gt;a<sub>0</sub> for that. We want
  70 both. And in possibly much more complex versions.
  71 </p>
  72 <p>
  73 Ideas:
  74 </p><ul>
  75 <li>
  76 I first thought of decorating the individual words/tags in a
  77 sentence with their indices, and perhaps just duplicating the
  78 relevant rules (one for each index of the duplicate tags). But this
  79 gives an explosion in attachment rules (although a contained
  80 explosion, within the rules used in a sentence; but most sentences
  81 will have at least two NN's so it will be a problem).
  82 </li>
  83 <li>
  84 Then, I had a <i>brilliant</i> idea. Just let e(), the helper function of
  85 inner(), parametrize for an extra pair of boolean values for whether
  86 or not we've attached anything to the left or right yet ("yet"
  87 meaning "below"). So now, e() has a chart of the form [s, t, LHS,
  88 Lattach, Rattach], and of course e(s,t,LHS) is the sum of the four
  89 possible values for (Lattach,Rattach). This makes e() lots more
  90 complex and DMV-specific though, so it's been rewritten in
  91 inner_dmv() in dmv.py.
  92 </li>
  93 <li>
  94 But, the problem is that we still need to be able get inner
  95 probabilites for only trees where the h at location loc_h is the
  96 root of the dependency. The previous idea does not give that. So I
  97 made <i>another</i> version of inner() that parametrizes for loc_h in the
  98 chart (instead of Lattach and Rattach). loc_h also lets us check
  99 whether we need probN or probA, in a much simpler fashion than the
 100 Lattach / Rattach version.
 101 </li>
 102 <li id="sec-2.1"><span class="todo">TODO</span> document this adjacency stuff better<br/>
 103 </li>
 104 <li id="sec-2.2"><span class="todo">TODO</span> [#A] test and debug my brilliant idea<br/>
 105 <ul>
 106 <li id="sec-2.2.1">Works on simple 'h h' sentence, but not with P_STOP, why?<br/>
 107 </li>
 108 </ul>
 109 </li>
 110 <li id="sec-2.3"><span class="done">DONE</span> implement my brilliant idea.<br/>
 111 <span class="timestamp-kwd">CLOSED: </span> <span class="timestamp">2008-06-01 Sun 17:19</span><br/>
 112 <a href="src/dmv.py">e(sti) in dmv.py</a>
 113
 114 </li>
 115 <li id="sec-2.4"><span class="done">DONE</span> [#A] test inner() on sentences with duplicate words<br/>
 116 Works with eg. the sentence "h h h"
 117
 118
 119 </li>
 120 </ul>
 121 </div>
 122
 123 </div>
 124
 125 <div id="outline-container-3" class="outline-2">
 126 <h2 id="sec-3">3 <span class="todo">TODO</span> [#A] P_STOP for IO/EM</h2>
 127 <div id="text-3">
 128
 129 <p><a href="src/dmv.py">dmv-P_STOP</a>
 130 Remember: The P<sub>STOP</sub> formula is upside-down (left-to-right also).
 131 (In the article..not the thesis)
 132 </p>
 133 <p>
 134 Remember: Initialization makes some "short-cut" rules, these will also
 135 have to be updated along with the other P<sub>STOP</sub> updates:
 136 </p><ul>
 137 <li>
 138 b[(NOBAR, n<sub>h</sub>), 'h'] = 1.0       # always
 139 </li>
 140 <li>
 141 b[(RBAR, n<sub>h</sub>), 'h'] = h_.probA  # h_ is RBAR stop rule
 142 </li>
 143 <li>
 144 b[(LRBAR, n<sub>h</sub>), 'h'] = h_.probA * _ h_.probA
 145
 146 </li>
 147 <li id="sec-3.1">How is the P_STOP formula different given other values for dir and adj?<br/>
 148 Assuming this:
 149 <ul>
 150 <li>
 151 P<sub>STOP</sub>(STOP|h,L,non_adj) = &sum;<sub>corpus</sub> &sum;<sub>s&lt;loc(h)</sub> &sum;<sub>t</sub>
 152 inner(s,t,(LRBAR,h)&hellip;) / &sum;<sub>corpus</sub> &sum;<sub>s&lt;loc(h)</sub> &sum;<sub>t</sub> inner(s,t,(RBAR,h)&hellip;)
 153 </li>
 154 <li>
 155 P<sub>STOP</sub>(STOP|h,L,adj) = &sum;<sub>corpus</sub> &sum;<sub>s=loc(h)</sub> &sum;<sub>t</sub>
 156 inner(s,t,(LRBAR,h)&hellip;) / &sum;<sub>corpus</sub> &sum;<sub>s=loc(h)</sub> &sum;<sub>t</sub> inner(s,t,(RBAR,h)&hellip;)
 157 </li>
 158 <li>
 159 P<sub>STOP</sub>(STOP|h,R,non_adj) = &sum;<sub>corpus</sub> &sum;<sub>s</sub> &sum;<sub>t&gt;loc(h)</sub>
 160 inner(s,t,(LRBAR,h)&hellip;) / &sum;<sub>corpus</sub> &sum;<sub>s</sub> &sum;<sub>t&gt;loc(h)</sub> inner(s,t,(RBAR,h)&hellip;)
 161 </li>
 162 <li>
 163 P<sub>STOP</sub>(STOP|h,R,adj) = &sum;<sub>corpus</sub> &sum;<sub>s</sub> &sum;<sub>t=loc(h)</sub>
 164 inner(s,t,(LRBAR,h)&hellip;) / &sum;<sub>corpus</sub> &sum;<sub>s</sub> &sum;<sub>t=loc(h)</sub> inner(s,t,(RBAR,h)&hellip;)
 165
 166
 167
 168 </li>
 169 </ul>
 170
 171 <p>(And P<sub>STOP</sub>(-STOP|&hellip;) = 1 - P<sub>STOP</sub>(STOP|&hellip;) )
 172 </p></li>
 173 </ul>
 174 </div>
 175
 176 </div>
 177
 178 <div id="outline-container-4" class="outline-2">
 179 <h2 id="sec-4">4 <span class="todo">TODO</span> P_CHOOSE for IO/EM</h2>
 180 <div id="text-4">
 181
 182 <p>Write the formulas! should be easy?
 183 </p></div>
 184
 185 </div>
 186
 187 <div id="outline-container-5" class="outline-2">
 188 <h2 id="sec-5">5 Initialization   </h2>
 189 <div id="text-5">
 190
 191 <p><a href="/Users/kiwibird/Documents/Skole/V08/Probability/dmvccm/src/dmv.py">dmv-inits</a>
 192 </p>
 193 <p>
 194 We do have to go through the corpus, since the probabilities are based
 195 on how far away in the sentence arguments are from their heads.
 196 </p><ul>
 197 <li id="sec-5.1"><span class="todo">TODO</span> Separate initialization to another file?                      &nbsp;&nbsp;&nbsp;<span class="tag">PRETTIER</span><br/>
 198 (It's rather messy.)
 199 </li>
 200 <li id="sec-5.2"><span class="todo">TOGROK</span> CCM Initialization    <br/>
 201 P<sub>SPLIT</sub> used here&hellip; how, again?
 202 </li>
 203 <li id="sec-5.3"><span class="done">DONE</span> DMV Initialization probabilities<br/>
 204 (from initialization frequency)
 205 </li>
 206 <li id="sec-5.4"><span class="done">DONE</span> DMV Initialization frequencies    <br/>
 207 <span class="timestamp-kwd">CLOSED: </span> <span class="timestamp">2008-05-27 Tue 20:04</span><br/>
 208 <ul>
 209 <li id="sec-5.4.1">P_STOP    <br/>
 210 P<sub>STOP</sub> is not well defined by K&amp;M. One possible interpretation given
 211 the sentence [det nn vb nn] is
 212 <pre>
 213  f_{STOP}( STOP|det, L, adj) +1
 214  f_{STOP}(-STOP|det, L, adj) +0
 215  f_{STOP}( STOP|det, L, non_adj) +1
 216  f_{STOP}(-STOP|det, L, non_adj) +0
 217  f_{STOP}( STOP|det, R, adj) +0
 218  f_{STOP}(-STOP|det, R, adj) +1
 219
 220  f_{STOP}( STOP|nn, L, adj) +0
 221  f_{STOP}(-STOP|nn, L, adj) +1
 222  f_{STOP}( STOP|nn, L, non_adj) +1  # since there's at least one to the left
 223  f_{STOP}(-STOP|nn, L, non_adj) +0
 224 </pre>
 225 <ul>
 226 <li id="sec-5.4.1.1"><span class="todo">TODO</span> tweak<br/>
 227 <pre>
 228             f[head,  'STOP', 'LN'] += (i_h &lt;= 1)     # first two words
 229             f[head, '-STOP', 'LN'] += (not i_h &lt;= 1)
 230             f[head,  'STOP', 'LA'] += (i_h == 0)     # very first word
 231             f[head, '-STOP', 'LA'] += (not i_h == 0)
 232             f[head,  'STOP', 'RN'] += (i_h &gt;= n - 2) # last two words
 233             f[head, '-STOP', 'RN'] += (not i_h &gt;= n - 2)
 234             f[head,  'STOP', 'RA'] += (i_h == n - 1) # very last word
 235             f[head, '-STOP', 'RA'] += (not i_h == n - 1)
 236 </pre>
 237 vs
 238 <pre>
 239             # this one requires some additional rewriting since it
 240             # introduces divisions by zero
 241             f[head,  'STOP', 'LN'] += (i_h == 1)     # second word
 242             f[head, '-STOP', 'LN'] += (not i_h &lt;= 1) # not first two
 243             f[head,  'STOP', 'LA'] += (i_h == 0)     # first word
 244             f[head, '-STOP', 'LA'] += (not i_h == 0) # not first
 245             f[head,  'STOP', 'RN'] += (i_h == n - 2)     # second-to-last
 246             f[head, '-STOP', 'RN'] += (not i_h &gt;= n - 2) # not last two
 247             f[head,  'STOP', 'RA'] += (i_h == n - 1)     # last word
 248             f[head, '-STOP', 'RA'] += (not i_h == n - 1) # not last
 249 </pre>
 250 vs
 251 <pre>
 252             f[head,  'STOP', 'LN'] += (i_h == 1)     # second word
 253             f[head, '-STOP', 'LN'] += (not i_h == 1) # not second
 254             f[head,  'STOP', 'LA'] += (i_h == 0)     # first word
 255             f[head, '-STOP', 'LA'] += (not i_h == 0) # not first
 256             f[head,  'STOP', 'RN'] += (i_h == n - 2)     # second-to-last
 257             f[head, '-STOP', 'RN'] += (not i_h == n - 2) # not second-to-last
 258             f[head,  'STOP', 'RA'] += (i_h == n - 1)     # last word
 259             f[head, '-STOP', 'RA'] += (not i_h == n - 1) # not last
 260 </pre>
 261 vs
 262 "all words take the same number of arguments" interpreted as
 263 <pre>
 264 for all heads:
 265     p_STOP(head, 'STOP', 'LN') = 0.3
 266     p_STOP(head, 'STOP', 'LA') = 0.5
 267     p_STOP(head, 'STOP', 'RN') = 0.4
 268     p_STOP(head, 'STOP', 'RA') = 0.7
 269 </pre>
 270 (which we easily may tweak in init_zeros())
 271 </li>
 272 </ul>
 273 </li>
 274 <li id="sec-5.4.2">P_CHOOSE<br/>
 275 Go through the corpus, counting distances between heads and
 276 arguments. In [det nn vb nn], we give
 277 <ul>
 278 <li>
 279 f<sub>CHOOSE</sub>(nn|det, R) +1/1 + C
 280 </li>
 281 <li>
 282 f<sub>CHOOSE</sub>(vb|det, R) +1/2 + C
 283 </li>
 284 <li>
 285 f<sub>CHOOSE</sub>(nn|det, R) +1/3 + C
 286 <ul>
 287 <li>
 288 If this were the full corpus, P<sub>CHOOSE</sub>(nn|det, R) would have
 289 (1+1/3+2C) / sum_a f<sub>CHOOSE</sub>(a|det, R)
 290
 291 </li>
 292 </ul>
 293 </li>
 294 </ul>
 295
 296 <p>The ROOT gets "each argument with equal probability", so in a sentence
 297 of three words, 1/3 for each (in [nn vb nn], 'nn' gets 2/3). Basically
 298 just a frequency count of the corpus&hellip;
 299 </p></li>
 300 </ul>
 301 </li>
 302 </ul>
 303 </div>
 304
 305 </div>
 306
 307 <div id="outline-container-6" class="outline-2">
 308 <h2 id="sec-6">6 [#C] Deferred</h2>
 309 <div id="text-6">
 310
 311 <ul>
 312 <li id="sec-6.1"><span class="todo">TODO</span> inner_dmv() should disregard rules with heads not in sent     &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 313 If the sentence is "nn vbd det nn", we should not even look at rules
 314 where
 315 <pre>
 316  rule.head() not in "nn vbd det nn".split()
 317 </pre>
 318 This is ruled out by getting rules from g.rules(LHS, sent).
 319
 320 <p>
 321 Also, we optimize this further by saying we don't even recurse into
 322 attachment rules where
 323 <pre>
 324  rule.head() not in sent[ s :r+1]
 325  rule.head() not in sent[r+1:t+1]
 326 </pre>
 327 meaning, if we're looking at the span "vbd det", we only use
 328 attachment rules where both daughters are members of ['vbd','det']
 329 (although we don't (yet) care about removing rules that rewrite to the
 330 same tag if there are no duplicate tags in the span, etc., that would
 331 be a lot of trouble for little potential gain).
 332 </p></li>
 333 <li id="sec-6.2"><span class="todo">TODO</span> when reestimating P_STOP etc, remove rules with p &lt; epsilon   &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 334 </li>
 335 <li id="sec-6.3"><span class="todo">TODO</span> inner_dmv, short ranges and impossible attachment             &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 336 If s-t &lt;= 2, there can be only one attachment below, so don't recurse
 337 with both Lattach=True and Rattach=True.
 338
 339 <p>
 340 If s-t &lt;= 1, there can be no attachment below, so only recurse with
 341 Lattach=False, Rattach=False.
 342 </p>
 343 <p>
 344 Put this in the loop under rewrite rules (could also do it in the STOP
 345 section, but that would only have an effect on very short sentences).
 346 </p></li>
 347 <li id="sec-6.4"><span class="todo">TODO</span> clean up the module files                                     &nbsp;&nbsp;&nbsp;<span class="tag">PRETTIER</span><br/>
 348 Is there better way to divide dmv and harmonic? There's a two-way
 349 dependency between the modules. Guess there could be a third file that
 350 imports both the initialization and the actual EM stuff, while a file
 351 containing constants and classes could be imported by all others:
 352 <pre>
 353  dmv.py imports dmv_EM.py imports dmv_classes.py
 354  dmv.py imports dmv_inits.py imports dmv_classes.py
 355 </pre>
 356
 357 </li>
 358 <li id="sec-6.5"><span class="todo">TOGROK</span> Some (tagged) sentences are bound to come twice             &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 359 Eg, first sort and count, so that the corpus
 360 [['nn','vbd','det','nn'],
 361 ['vbd','nn','det','nn'],
 362 ['nn','vbd','det','nn']]
 363 becomes
 364 [(['nn','vbd','det','nn'],2),
 365 (['vbd','nn','det','nn'],1)]
 366 and then in each loop through sentences, make sure we handle the
 367 frequency correctly.
 368
 369 <p>
 370 Is there much to gain here?
 371 </p>
 372 </li>
 373 <li id="sec-6.6"><span class="todo">TOGROK</span> tags as numbers or tags as strings?                         &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 374 Need to clean up the representation.
 375
 376 <p>
 377 Stick with tag-strings in initialization then switch to numbers for
 378 IO-algorithm perhaps? Can probably afford more string-matching in
 379 initialization..
 380 </p></li>
 381 </ul>
 382 </div>
 383
 384 </div>
 385
 386 <div id="outline-container-7" class="outline-2">
 387 <h2 id="sec-7">7 Expectation Maximation in IO/DMV-terms</h2>
 388 <div id="text-7">
 389
 390 <p>inner(s,t,LHS) calculates the expected number of trees headed by LHS
 391 from s to t (sentence positions). This uses the P_STOP and P_CHOOSE
 392 values, which have been conveniently distributed into CNF rules as
 393 probN and probA (non-adjacent and adjacent probabilites).
 394 </p>
 395 <p>
 396 When re-estimating, we use the expected values from inner() to get new
 397 values for P_STOP and P_CHOOSE. When we've re-estimated for the entire
 398 corpus, we distribute P_STOP and P_CHOOSE into the CNF rules again, so
 399 that in the next round we use new probN and probA to find
 400 inner-probabilites.
 401 </p>
 402 <p>
 403 The distribution of P_STOP and P_CHOOSE into CNF rules also happens in
 404 init_normalize() (here along with the creation of P_STOP and
 405 P_CHOOSE); P_STOP is used to create CNF rules where one branch of the
 406 rule is STOP, P_CHOOSE is used to create rules of the form
 407 <pre>
 408  h  -&gt; h  _a_
 409  h_ -&gt; h_ _a_
 410 </pre>
 411 </p>
 412 <p>
 413 Since "adjacency" is not captured in regular CNF rules, we need two
 414 probabilites for each rule, and inner() has to know when to use which.
 415 </p>
 416 <ul>
 417 <li id="sec-7.1"><span class="todo">TODO</span> Corpus access<br/>
 418 </li>
 419 <li id="sec-7.2"><span class="todo">TOGROK</span> sentences or rules as the "outer loop"?                     &nbsp;&nbsp;&nbsp;<span class="tag">OPTIMIZE</span><br/>
 420 In regard to the E/M-step, finding P<sub>STOP</sub>, P<sub>CHOOSE</sub>.
 421
 422
 423 </li>
 424 </ul>
 425 </div>
 426
 427 </div>
 428
 429 <div id="outline-container-8" class="outline-2">
 430 <h2 id="sec-8">8 Python-stuff</h2>
 431 <div id="text-8">
 432
 433 <ul>
 434 <li>
 435 <a href="src/pseudo.py">pseudo.py</a>
 436 </li>
 437 <li>
 438 <a href="http://nltk.org/doc/en/structured-programming.html">http://nltk.org/doc/en/structured-programming.html</a> recursive dynamic
 439 </li>
 440 <li>
 441 <a href="http://nltk.org/doc/en/advanced-parsing.html">http://nltk.org/doc/en/advanced-parsing.html</a>
 442 </li>
 443 <li>
 444 <a href="http://jaynes.colorado.edu/PythonIdioms.html">http://jaynes.colorado.edu/PythonIdioms.html</a>
 445
 446
 447
 448 </li>
 449 </ul>
 450 </div>
 451
 452 </div>
 453
 454 <div id="outline-container-9" class="outline-2">
 455 <h2 id="sec-9">9 Git</h2>
 456 <div id="text-9">
 457
 458 <p>Setting up a new project:
 459 <pre>
 460  git init
 461  git add .
 462  git commit -m "first release"
 463 </pre>
 464 </p>
 465 <p>
 466 Later on: (<code>-a</code> does <code>git rm</code> and <code>git add</code> automatically)
 467 <pre>
 468  git init
 469  git commit -a -m "some subsequent release"
 470 </pre>
 471 </p>
 472 <p>
 473 Then push stuff up to the remote server:
 474 <pre>
 475  git push git+ssh://username@repo.or.cz/srv/git/dmvccm.git master
 476 </pre>
 477 </p>
 478 <p>
 479 (<code>eval `ssh-agent`</code> and <code>ssh-add</code> to avoid having to type in keyphrase all
 480 the time)
 481 </p>
 482 <p>
 483 Make a copy of the (remote) master branch:
 484 <pre>
 485  git clone git://repo.or.cz/dmvccm.git
 486 </pre>
 487 </p>
 488 <p>
 489 Make and name a new branch in this folder
 490 <pre>
 491  git checkout -b mybranch
 492 </pre>
 493 </p>
 494 <p>
 495 To save changes in <code>mybranch</code>:
 496 <pre>
 497  git commit -a
 498 </pre>
 499 </p>
 500 <p>
 501 Go back to the master branch (uncommitted changes from <code>mybranch</code> are
 502 carried over):
 503 <pre>
 504  git checkout master
 505 </pre>
 506 </p>
 507 <p>
 508 Try out:
 509 <pre>
 510  git add --interactive
 511 </pre>
 512 </p>
 513 <p>
 514 Good tutorial:
 515 <a href="http://www-cs-students.stanford.edu/~blynn//gitmagic/">http://www-cs-students.stanford.edu/~blynn//gitmagic/</a>
 516 </p></div>
 517 </div>
 518 <div id="postamble"><p class="author"> Author: Kevin Brubeck Unhammer
 519 <a href="mailto:K.BrubeckUnhammer at student uva nl ">&lt;K.BrubeckUnhammer at student uva nl &gt;</a>
 520 </p>
 521 <p class="date"> Date: 2008/06/06 19:12:56</p>
 522 </div><p class="postamble">Skrive vha. emacs + <a href='http://orgmode.org/'>org-mode</a></p></body>
 523 </html>