File size: 26,416 Bytes
69fb171 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<link rev="made" href="mailto:[email protected]">
<title>CRF++: Yet Another CRF toolkit</title>
<link type="text/css" rel="stylesheet" href="default.css">
</head>
<body>
<h1>CRF++: Yet Another CRF toolkit</h1>
<h2>Introduction</h2>
<p><b>CRF++</b> is a simple, customizable, and open source
implementation of <a href="http://www.cis.upenn.edu/~pereira/papers/crf.pdf">Conditional Random Fields (CRFs)</a>
for segmenting/labeling sequential data. CRF++ is designed for generic purpose and will be applied to a variety of NLP tasks, such as
Named Entity Recognition, Information Extraction and Text Chunking.
<h2>Table of contents</h2>
<ul>
<li><a href="#features">Features</a></li>
<li><a href="#news">News</a></li>
<li><a href="#download">Download</a> </li>
<ul>
<li><a href="#source">Source</a></li>
<li><a href="#windows">Binary package for MS-Windows</a></li>
</ul>
<li><a href="#install">Installation</a></li>
<li>
<a href="#usage">Usage</a>
<ul>
<li><a href="#format">Training and Test file formats</a></li>
<li><a href="#templ">Preparing feature templates</a></li>
<li><a href="#training">Training (encoding)</a></li>
<li><a href="#testing">Testing (decoding)</a></li>
</ul>
</li>
<li><a href="#tips">Case studies</a></li>
<li><a href="#tips">Useful Tips</a></li>
<li><a href="#todo">To do</a></li>
<li><a href="#links">Links</a></li>
</ul>
<h2><a name="features">Features</a></h2>
<ul>
<li>Can redefine feature sets</li>
<li>Written in C++ with STL</li>
<li>Fast training based on <a href="http://www-fp.mcs.anl.gov/otc/Guide/SoftwareGuide/Blurbs/lbfgs.html">LBFGS</a>, a quasi-newton algorithm
for large scale numerical optimization problem</li>
<li>Less memory usage both in training and testing</li>
<li>encoding/decoding in practical time</li>
<li>Can perform n-best outputs</li>
<li>Can perform single-best MIRA training</li>
<li>Can output marginal probabilities for all candidates</li>
<li>Available as an open source software</li>
</ul>
<h2><a name="news">News</a></h2>
<ul>
<strong>2013-02-13</strong>: <a href="#download">CRF++ 0.58</a> Released<br>
<ul>
<li>Added createModelFromArray() method to load model file from fixed buffer.
<li>Added getTemplate() method to get template string.
</ul>
<strong>2012-03-25</strong><br>
<ul>
<li>Fixed build issue around libtool.
<li>Fixed C++11 compatible issue.
</ul>
<strong>2012-02-24</strong><br>
<ul>
<li>Added CRFPP:Tagger::set_model() method.
<li>Fixed minor bugs
</ul>
<strong>2012-02-15</strong>: CRF++ 0.55<br>
<ul>
<li>Added new CRFPP:Model class so that multiple threads can share single CRF++ model.
<li>Added Tagger::set_penalty and Tagger::penalty() method for dual decompositon decoding
<li>Fixed crash bug on Windows
<li>Fixed minor bugs
</ul>
<strong>2010-05-16</strong>: CRF++ 0.54</a> Released<br>
<ul>
<li>fixed the bug in L1 regularization. Reported by Fujii Yasuhisa</li>
</ul>
<strong>2009-05-06</strong>: CRF++ 0.5 Released<br>
<ul>
<li>fixed build failure on libtool
</ul>
<strong>2009-04-19</strong>: CRF++ 0.52<br>
<ul>
<li>Code clean up
<li>replaced obsolete sstream with stringstream
</ul>
<strong>2007-07-12</strong>: CRF++ 0.51</a><br>
<ul>
<li>Fixed a compilation error on gcc 4.3
</ul>
<strong>2007-12-09</strong>: CRF++ 0.50<br>
<ul>
<li>Bug fix in --convert mode (Could not generate model from text file)
</ul>
<strong>2007-08-18</strong>: CRF++ 0.49<br>
<ul>
<li>Added setter/getter for nbest, cost_factor and vlevel to API
</ul>
<strong>2007-07-07</strong>: CRF++ 0.48 Released<br>
<ul>
<li>Support L1-CRF. use -a CRF-L1 option to enable L1 regularization.
</ul>
<strong>2007-03-07</strong>: CRF++ 0.47 Released<br>
<ul>
<li>Fixed a bug in MIRA training
</ul>
<strong>2007-02-12</strong>: CRF++ 0.46 Released<br>
<ul>
<li>Changed the licence from LGPL to LGPL/BSD dual
license
<li>Perl/Ruby/Python/Java binding supports (see
perl/ruby/python/java directory respectively)
<li>Code refactoring
</ul>
<strong>2006-11-26</strong>: CRF++ 0.45<br>
<ul>
<li>Support 1-best MIRA training (use -a MIRA option)
</ul>
<strong>2006-08-18</strong>: CRF++ 0.44<br>
<ul>
<li>Fixed a bug in feature extraction</li>
<li>Allowed redundant spaces in training/test files</li>
<li>Determined real column size by looking at template</li>
<li>Added sample code of API (sdk/example.cpp)
<li>Described usage of each API function (crfpp.h)
</ul>
<strong>2006-08-07</strong>: CRF++ 0.43<br>
<ul>
<li>implemented several API functions to get lattice
information</li>
<li>added -c option to control cost-factor
</ul>
<strong>2006-03-31</strong>: CRF++ 0.42<br>
<ul>
<li>Fixed a bug in feature extraction</li>
</ul>
<strong>2006-03-30</strong>: CRF++ 0.41<br>
<ul>
<li>Support parallel training</li>
</ul>
<strong>2006-03-21</strong>: CRF++ 0.40<br>
<ul>
<li>Fixed a fatal memory leak bug</li>
<li>make CRF++ API</li>
</ul>
<strong>2005-10-29</strong>: CRF++ 0.3</a>
<ul>
<li>added -t option that enables you to have not only binary
model but also text model
<li>added -C option for converting a text model to a binary model
</ul>
<strong>2005-07-04</strong>: CRF++ 0.2
Released<br>
<ul>
<li>Fixed several bugs</li>
</ul>
<strong>2005-05-28</strong>: CRF++ 0.1
Released<br>
<ul>
<li>Initial Release</li>
</ul>
</ul>
<h2><a name="download">Download</a></h2>
<ul>
<li><b>CRF++</b> is free software; you can redistribute it
and/or modify it under the terms of the <a href=
"http://www.gnu.org/copyleft/lesser.html">GNU Lesser General
Public License</a> or <a
href="http://www.opensource.org/licenses/bsd-license.php">new BSD License</a></li>
<li>
Please let <a href=
"mailto:[email protected]">me</a> know if you use
<b>CRF++</b> for research purpose or find any research
publications where <b>CRF++</b> is applied.
<h3><a name="source">Source</a></h3>
<ul>
<li>CRF++-0.58.tar.gz: <a href="http://code.google.com/p/crfpp/downloads/list">HTTP</a></li>
</ul>
<h3><a name="windows">Binary package for MS-Windows</a></h3>
<ul>
<li><a href="http://code.google.com/p/crfpp/downloads/list">HTTP</a><br>
</ul>
</li>
</ul>
<h2><a name="install">Installation</a></h2>
<ul>
<li>
Requirements
<ul>
<li>C++ compiler (gcc 3.0 or higher)</li>
</ul>
</li>
<li>
How to make
<pre>
% ./configure
% make
% su
# make install
</pre>
You can change default install path by using --prefix
option of configure script.<br>
Try --help option for finding out other options.
</li>
</ul>
<h2><a name="usage">Usage</a></h2>
<h3><a name="format">Training and Test file formats</a></h3>
<p>Both the training file and the test file need to be in a
particular format for <b>CRF++</b> to work properly.
Generally speaking, training and test file must consist of
multiple <b>tokens</b>. In addition, a <b>token</b>
consists of multiple (but fixed-numbers) columns. The
definition of tokens depends on tasks, however, in
most of typical cases, they simply correspond to
<b>words</b>. Each token must be represented in one line,
with the columns separated by white space (spaces or
tabular characters). A sequence of token becomes a
<b>sentence</b>. To identify the boundary between
sentences, an empty line is put.</p>
<p>You can give as many columns as you like, however the
number of columns must be fixed through all tokens.
Furthermore, there are some kinds of "semantics" among the
columns. For example, 1st column is 'word', second column
is 'POS tag' third column is 'sub-category of POS' and so
on.</p>
<p>The last column represents a true answer tag which is going
to be trained by CRF.</p>
<p>Here's an example of such a file: (data for CoNLL shared
task)</p>
<pre>
He PRP B-NP
reckons VBZ B-VP
the DT B-NP
current JJ I-NP
account NN I-NP
deficit NN I-NP
will MD B-VP
narrow VB I-VP
to TO B-PP
only RB B-NP
# # I-NP
1.8 CD I-NP
billion CD I-NP
in IN B-PP
September NNP B-NP
. . O
He PRP B-NP
reckons VBZ B-VP
..
</pre>
<p>There are 3 columns for each token.</p>
<ul>
<li>The word itself (e.g. reckons);</li>
<li>part-of-speech associated with the word (e.g. VBZ);</li>
<li>Chunk(answer) tag represented in IOB2 format;</li>
</ul>
<p>The following data is invalid, since the number of
columns of second and third are 2. (They have no POS
column.) The number of columns should be fixed.</p>
<pre>
He PRP B-NP
reckons B-VP
the B-NP
current JJ I-NP
account NN I-NP
..
</pre>
<h3><a name="templ">Preparing feature templates</a></h3>
<p>
As CRF++ is designed as a general purpose tool, you have to
specify the feature templates in advance. This file describes
which features are used in training and testing.
</p>
<ul>
<li>Template basic and macro</li>
<p>
Each line in the template file denotes one <i>template</i>.
In each template, special macro <i>%x[row,col]</i> will be
used to specify a token in the input data. <i>row</i> specfies the
relative position from the current focusing token
and <i>col</i> specifies the absolute position of the column.
</p>
<p>Here you can find some examples for the replacements</p>
<pre>
Input: Data
He PRP B-NP
reckons VBZ B-VP
the DT B-NP << CURRENT TOKEN
current JJ I-NP
account NN I-NP
</pre>
<p>
<table border>
<tr>
<td>template</td>
<td>expanded feature</td>
</tr>
<tr>
<td><b>%x[0,0]</b></td>
<td>the</td>
</tr>
<tr>
<td><b>%x[0,1]</b></td>
<td>DT</td>
</tr>
<tr>
<td><b>%x[-1,0]</b></td>
<td>rokens</td>
</tr>
<tr>
<td><b>%x[-2,1]</b></td>
<td>PRP</td>
</tr>
<tr>
<td><b>%x[0,0]/%x[0,1]</b></td>
<td>the/DT</td>
</tr>
<tr>
<td><b>ABC%x[0,1]123</b></td>
<td>ABCDT123</td>
</tr>
</table>
</p>
<br>
<li>Template type</li>
<p>Note also that there are two types of templates.
The types are specified with the first character of templates.
</p>
<ul>
<li>Unigram template: first character, <b>'U'</b></li>
<p>
This is a template to describe unigram features.
When you give a template "U01:%x[0,1]", CRF++ automatically
generates a set of feature functions (func1 ... funcN) like:
</p>
<pre>
func1 = if (output = B-NP and feature="U01:DT") return 1 else return 0
func2 = if (output = I-NP and feature="U01:DT") return 1 else return 0
func3 = if (output = O and feature="U01:DT") return 1 else return 0
....
funcXX = if (output = B-NP and feature="U01:NN") return 1 else return 0
funcXY = if (output = O and feature="U01:NN") return 1 else return 0
...</pre>
<p>
The number of feature functions generated by a template amounts to
(L * N), where L is the number of output classes and N is the
number of unique string expanded from the given template.
</p>
<li>Bigram template: first character, <b>'B'</b></li>
<p>
This is a template to describe bigram features.
With this template, a combination of the current output token and previous output token
(bigram) is automatically generated. Note that this type of template generates a total of
(L * L * N) distinct features, where L is the
number of output classes and N is the number
of unique features generated by the templates.
When the number of classes is large, this type of templates would produce
a tons of distinct features that would cause inefficiency both
in training/testing.
</p>
<li>What is the diffrence between unigram and bigram features?</li>
<p>
The words unigram/bigram are confusing, since a macro for unigram-features
does allow you to write word-level bigram like %x[-1,0]%x[0,0]. Here,
unigram and bigram features mean uni/bigrams of output tags.</p>
<ul>
<li>unigram: |output tag| x |all possible strings expanded with a macro|</li>
<li>bigram: |output tag| x |output tag| x |all possible strings expanded with a macro|</li>
</ul>
<p></p>
</ul>
<li>Identifiers for distinguishing relative positions</li>
<p>
You also need to put an identifier in templates when relative positions of
tokens must be distinguished.
</p>
<p>
In the following case, the macro "%x[-2,1]" and "%x[1,1]" will be replaced
into "DT". But they indicates different "DT".
</p>
<pre>
The DT B-NP
pen NN I-NP
is VB B-VP << CURRENT TOKEN
a DT B-NP
</pre>
<p>To distinguish both two, put an unique identifier (U01: or U02:) in the
template:</p>
<pre>
U01:%x[-2,1]
U02:%x[1,1]
</pre>
<p>
In this case both two templates are regarded as different ones, as
they are expanded into different features, "U01:DT" and "U02:DT".
You can use any identifier whatever you like, but
it is useful to use numerical numbers to manage them, because they simply
correspond to feature IDs.
</p>
<p>
If you want to use "bag-of-words" feature, in other words,
not to care the relative position of features, You don't need to
put such identifiers.
</p>
<li>Example</li>
<p>Here is the template example for <a href="http://www.cnts.ua.ac.be/conll2000/chunking/">CoNLL 2000</a> shared task and Base-NP chunking
task. Only one bigram template ('B') is used. This means that
only combinations of previous output token and current token are
used as bigram features. The lines starting from # or empty lines are
discarded as comments</p>
<pre>
# Unigram
U00:%x[-2,0]
U01:%x[-1,0]
U02:%x[0,0]
U03:%x[1,0]
U04:%x[2,0]
U05:%x[-1,0]/%x[0,0]
U06:%x[0,0]/%x[1,0]
U10:%x[-2,1]
U11:%x[-1,1]
U12:%x[0,1]q
U13:%x[1,1]
U14:%x[2,1]
U15:%x[-2,1]/%x[-1,1]
U16:%x[-1,1]/%x[0,1]
U17:%x[0,1]/%x[1,1]
U18:%x[1,1]/%x[2,1]
U20:%x[-2,1]/%x[-1,1]/%x[0,1]
U21:%x[-1,1]/%x[0,1]/%x[1,1]
U22:%x[0,1]/%x[1,1]/%x[2,1]
# Bigram
B
</pre>
</ul>
</ul>
<h3><a name="training">Training (encoding)</a></h3>
<p>Use <i>crf_learn</i> command:
<pre>
% crf_learn template_file train_file model_file
</pre>
<p>
where <i>template_file</i> and <i>train_file</i>
are the files you need to prepare in advance.
<i>crf_learn</i> generates the trained model file in
<i>model_file</i>.
</p>
<p>crf_learn outputs the following information.</p>
<pre>
CRF++: Yet Another CRF Tool Kit
Copyright(C) 2005 Taku Kudo, All rights reserved.
reading training data: 100.. 200.. 300.. 400.. 500.. 600.. 700.. 800..
Done! 1.94 s
Number of sentences: 823
Number of features: 1075862
Number of thread(s): 1
Freq: 1
eta: 0.00010
C: 1.00000
shrinking size: 20
Algorithm: CRF
iter=0 terr=0.99103 serr=1.00000 obj=54318.36623 diff=1.00000
iter=1 terr=0.35260 serr=0.98177 obj=44996.53537 diff=0.17161
iter=2 terr=0.35260 serr=0.98177 obj=21032.70195 diff=0.53257
iter=3 terr=0.23879 serr=0.94532 obj=13642.32067 diff=0.35138
iter=4 terr=0.15324 serr=0.88700 obj=8985.70071 diff=0.34134
iter=5 terr=0.11605 serr=0.80680 obj=7118.89846 diff=0.20775
iter=6 terr=0.09305 serr=0.72175 obj=5531.31015 diff=0.22301
iter=7 terr=0.08132 serr=0.68408 obj=4618.24644 diff=0.16507
iter=8 terr=0.06228 serr=0.59174 obj=3742.93171 diff=0.18953
</pre>
<ul>
<li>iter: number of iterations processed</li>
<li>terr: error rate with respect to tags. (# of error tags/# of all tag)</li>
<li>serr: error rate with respect to sentences. (# of error sentences/#
of all sentences)</li>
<li>obj: current object value. When this value converges to a
fixed point, CRF++ stops the iteration.</li>
<li>diff: relative difference from the previous object value.</li>
</ul>
<p>
There are 4 major parameters to control the training condition
<ul>
<li>-a CRF-L2 or CRF-L1:<br>
Changing the regularization algorithm. Default setting is L2.
Generally speaking, L2 performs slightly better than L1, while
the number of non-zero features in L1 is drastically smaller than
that in L2.
<li>-c float: <br>
With this option, you can change the hyper-parameter for the CRFs.
With larger C value, CRF tends to overfit to the give training corpus.
This parameter trades the balance between overfitting and
underfitting. The results will significantly be influenced by
this parameter. You can find an optimal value by using
held-out data or more general model selection method such as
cross validation.
<li>-f NUM:<br>
This parameter sets the cut-off threshold for the features.
CRF++ uses the features that occurs no less than NUM times
in the given training data. The default value is 1.
When you apply CRF++ to large data, the number of unique features
would amount to several millions. This option is useful in such cases.
<li>-p NUM:<br>
If the PC has multiple CPUs, you can make the training faster
by using multi-threading. NUM is the number of threads.
</ul>
<p>Here is the example where these two parameters are used.</p>
<pre>
% crf_learn -f 3 -c 1.5 template_file train_file model_file
</pre>
<p>Since version 0.45, CRF++ supports single-best MIRA training.
MIRA training is used when -a MIRA option is set.
<pre>
% crf_learn -a MIRA template train.data model
CRF++: Yet Another CRF Tool Kit
Copyright(C) 2005 Taku Kudo, All rights reserved.
reading training data: 100.. 200.. 300.. 400.. 500.. 600.. 700.. 800..
Done! 1.92 s
Number of sentences: 823
Number of features: 1075862
Number of thread(s): 1
Freq: 1
eta: 0.00010
C: 1.00000
shrinking size: 20
Algorithm: MIRA
iter=0 terr=0.11381 serr=0.74605 act=823 uact=0 obj=24.13498 kkt=28.00000
iter=1 terr=0.04710 serr=0.49818 act=823 uact=0 obj=35.42289 kkt=7.60929
iter=2 terr=0.02352 serr=0.30741 act=823 uact=0 obj=41.86775 kkt=5.74464
iter=3 terr=0.01836 serr=0.25881 act=823 uact=0 obj=47.29565 kkt=6.64895
iter=4 terr=0.01106 serr=0.17011 act=823 uact=0 obj=50.68792 kkt=3.81902
iter=5 terr=0.00610 serr=0.10085 act=823 uact=0 obj=52.58096 kkt=3.98915
iter=0 terr=0.11381 serr=0.74605 act=823 uact=0 obj=24.13498 kkt=28.00000
...
</pre>
<ul>
<li>iter, terr, serror: same as CRF training</li>
<li>act: number of active examples in working set</li>
<li>uact: number of examples whose dual parameters reach soft margin
upper-bound C. 0 uact suggests that given training data was
linear separable</li>
<li>obj: current object value, ||w||^2</li>
<li>kkt: max kkt violation value. When it gets 0.0, MIRA training finishes</li>
</ul>
<p>There are some parameters to control the MIRA training condition</p>
<ul>
<li>-c float: <br>
Changes soft margin parameter, which is an analogue to the soft margin
parameter C in Support Vector Machines.
The definition is basically the same as -c option in CRF training.
With larger C value, MIRA tends to overfit to the give training
corpus.
<li>-f NUM:<br>
Same as CRF
<li>-H NUM:<br>
Changes shrinking size. When a training sentence is not used
in updating parameter vector NUM times, we can consider that the
instance doesn't contribute training any more. MIRA tries to
remove such instances. The process is called
"shrinking". When setting smaller NUM, shrinking occurs in early
stage, which drastically reduces
training time. However, too small NUM is not recommended.
When training finishes, MIRA tries to go through all training
examples again to know whether or not all KKT conditions are really
satisfied. Too small NUM would increase the chances of recheck.
</ul>
<h3><a name="testing">Testing (decoding)</a></h3>
<p>Use <i>crf_test</i> command:
<pre>
% crf_test -m model_file test_files ...
</pre>
<p>
where <i>model_file</i> is the file <i>crf_learn</i>creates.
In the testing, you don't need to specify the template file,
because the model file has the same information for the template.
<i>test_file</i> is the test data you want to assign sequential tags.
This file has to be written in the same format as training file.
</p>
<p>
Here is an output of <i>crf_test</i>:</p>
<pre>
% crf_test -m model test.data
Rockwell NNP B B
International NNP I I
Corp. NNP I I
's POS B B
Tulsa NNP I I
unit NN I I
..
</pre>
<p>The last column is given (estimated) tag.
If the 3rd column is true answer tag , you can evaluate the accuracy
by simply seeing the difference between the 3rd and 4th columns.</p>
<ul>
<li>verbose level</li>
<p>The <b>-v</b> option sets verbose level. default
value is 0. By increasing the level, you can have an
extra information from CRF++</p>
<ul>
<li>level 1 <br>
You can also have marginal probabilities for each tag
(a kind of confidece measure for each output tag)
and a conditional probably for the output (confidence measure for
the entire output).
<pre>
% crf_test -v1 -m model test.data| head
# 0.478113
Rockwell NNP B B/0.992465
International NNP I I/0.979089
Corp. NNP I I/0.954883
's POS B B/0.986396
Tulsa NNP I I/0.991966
...
</pre>
<p>
The first line "# 0.478113" shows the conditional probably for the output.
Also, each output tag has a probability represented like "B/0.992465".
</p>
<li>level 2<br>
<p>You can also have marginal probabilities for all other candidates.</p>
<pre>
% crf_test -v2 -m model test.data
# 0.478113
Rockwell NNP B B/0.992465 B/0.992465 I/0.00144946 O/0.00608594
International NNP I I/0.979089 B/0.0105273 I/0.979089 O/0.0103833
Corp. NNP I I/0.954883 B/0.00477976 I/0.954883 O/0.040337
's POS B B/0.986396 B/0.986396 I/0.00655976 O/0.00704426
Tulsa NNP I I/0.991966 B/0.00787494 I/0.991966 O/0.00015949
unit NN I I/0.996169 B/0.00283111 I/0.996169 O/0.000999975
..
</pre>
</ul>
<li>N-best outputs</li>
<p>
With the <b>-n</b> option, you can obtain N-best results
sorted by the conditional probability of CRF.
With n-best output mode, CRF++ first gives one additional line like "# N prob", where N means that
rank of the output starting from 0 and prob denotes the conditional
probability for the output. </p>
<p>Note that CRF++ sometimes
discards enumerating N-best results if it cannot find candidates any
more. This is the case when you give CRF++ a short
sentence.</p>
<p>CRF++ uses a combination of forward Viterbi and backward A* search. This combination
yields the exact list of n-best results. </p>
<p>Here is the example of the N-best results. </p>
<pre>
% crf_test -n 20 -m model test.data
# 0 0.478113
Rockwell NNP B B
International NNP I I
Corp. NNP I I
's POS B B
...
# 1 0.194335
Rockwell NNP B B
International NNP I I
</pre>
</ul>
</ul>
<h2><a name="testing">Tips</a></h2>
<ul>
<li>CRF++ uses the exactly same data format as <a
href="http://chasen.org/~taku/software/yamcha/">YamCha</a> uses.
You may use both two toolkits for an input data and compare the
performance between CRF and SVM
<li>The output of CRF++ is also compatible to <a href="http://www.cnts.ua.ac.be/conll2000/chunking/">CoNLL 2000</a> shared task.
This allows us to use the perl script
<a href="http://www.cnts.ua.ac.be/conll2000/chunking/output.html">
conlleval.pl</a> to
evaluate system outputs. This script is very useful and
give us a list of F-measures for all chunk types
</ul>
<h2><a name="training">Case studies</a></h2>
<p>
In the example directories, you can find three case studies, baseNP
chunking, Text Chunking, and Japanese named entity recognition, to use CRF++.
</p>
<p>
In each directory, please try the following commands
</p>
<pre> % crf_learn template train model
% crf_test -m model test </pre>
<h2><a name="todo">To Do</a></h2>
<ul>
<li>Support <a
href="http://www-2.cs.cmu.edu/~wcohen/postscript/semiCRF.pdf">semi-Markov
CRF</a>
<li>Support <a
href="http://www.cs.umass.edu/~mccallum/papers/lcrf-nips2004.pdf">
piece-wise CRF</a>
<li>Provide useful C++/C API (Currently no APIs are available)
</ul>
<h2><a name="links">References</a></h2>
<ul>
<li>J. Lafferty, A. McCallum, and F. Pereira.
<a href="http://www.cis.upenn.edu/~pereira/papers/crf.pdf">Conditional random fields: Probabilistic models for segmenting and
labeling sequence data</a>, In Proc. of ICML, pp.282-289, 2001
<li>F. Sha and F. Pereira. <a
href="http://www.cis.upenn.edu/~feisha/pubs/shallow03.pdf">Shallow
parsing with conditional random fields</a>, In Proc. of HLT/NAACL 2003
<li><a
href="http://staff.science.uva.nl/~erikt/research/np-chunking.html">NP chunking</a></li>
<li><a href= "http://www.cnts.ua.ac.be/conll2000/chunking/">CoNLL
2000 shared task: Chunking</a></li>
</ul>
<hr>
<p>$Id: index.html,v 1.23 2003/01/06 13:11:21 taku-ku Exp
$;</p>
<address>
[email protected]
</address>
</body>
</html>
|