crfpp_8h-source.html

CRF工具包

<a name="l00091"></a>00091   CRFPP_DLL_EXTERN <span class="keywordtype">float</span> <a class="code" href="crfpp_8h.html#e846df1fd9e207c4fca788c9bc17f1e9">crfpp_cost_factor</a>(<a class="code" href="crfpp_8h.html#513ea0b9519c24b6c1797d74c2202e51">crfpp_t</a> *);
<a name="l00092"></a>00092   CRFPP_DLL_EXTERN <span class="keywordtype">void</span> <a class="code" href="crfpp_8h.html#1122f31df4d3673108e0757353c38c01">crfpp_set_nbest</a>(<a class="code" href="crfpp_8h.html#513ea0b9519c24b6c1797d74c2202e51">crfpp_t</a> *, <span class="keywordtype">size_t</span>);
<a name="l00093"></a>00093 <span class="preprocessor">#endif</span>
<a name="l00094"></a>00094 <span class="preprocessor"></span>
<a name="l00095"></a>00095 <span class="preprocessor">#ifdef __cplusplus</span>
<a name="l00096"></a>00096 <span class="preprocessor"></span>}
<a name="l00097"></a>00097 <span class="preprocessor">#endif</span>
<a name="l00098"></a>00098 <span class="preprocessor"></span>
<a name="l00099"></a>00099 <span class="comment">/* C++ interface */</span>
<a name="l00100"></a>00100 <span class="preprocessor">#ifdef __cplusplus</span>
<a name="l00101"></a>00101 <span class="preprocessor"></span>
<a name="l00102"></a><a class="code" href="namespaceCRFPP.html">00102</a> <span class="keyword">namespace </span>CRFPP {
<a name="l00103"></a>00103 
<a name="l00104"></a><a class="code" href="classCRFPP_1_1Tagger.html">00104</a> <span class="keyword">class </span><a class="code" href="classCRFPP_1_1Tagger.html">Tagger</a> {
<a name="l00105"></a>00105  <span class="keyword">public</span>:
<a name="l00106"></a>00106 <span class="preprocessor">#ifndef SWIG</span>
<a name="l00107"></a>00107 <span class="preprocessor"></span>  <span class="comment">// open model with parameters in argv[]</span>
<a name="l00108"></a>00108   <span class="comment">// e.g, argv[] = {"CRF++", "-m", "model", "-v3"};</span>
<a name="l00109"></a>00109   <span class="keyword">virtual</span> <span class="keywordtype">bool</span> <a class="code" href="classCRFPP_1_1Tagger.html#64a7abe75bc6d41df69ad1862c3d73a0">open</a>(<span class="keywordtype">int</span> argc,  <span class="keywordtype">char</span>** argv) = 0;
<a name="l00110"></a>00110 
<a name="l00111"></a>00111   <span class="comment">// open model with parameter arg, e.g. arg = "-m model -v3";</span>
<a name="l00112"></a>00112   <span class="keyword">virtual</span> <span class="keywordtype">bool</span> <a class="code" href="classCRFPP_1_1Tagger.html#64a7abe75bc6d41df69ad1862c3d73a0">open</a>(<span class="keyword">const</span> <span class="keywordtype">char</span>* arg) = 0;
<a name="l00113"></a>00113 
<a name="l00114"></a>00114   <span class="comment">// add str[] as tokens to the current context</span>
<a name="l00115"></a>00115   <span class="keyword">virtual</span> <span class="keywordtype">bool</span> <span class="keyword">add</span>(<span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#0bb307faa0e68f71ca1bb6a992bc01e4">size</a>, <span class="keyword">const</span> <span class="keywordtype">char</span> **str) = 0;
<a name="l00116"></a>00116 
<a name="l00117"></a>00117   <span class="comment">// close the current model</span>
<a name="l00118"></a>00118   <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classCRFPP_1_1Tagger.html#17d23a6de9ae9dac536c98cca32a7f01">close</a>() = 0;
<a name="l00119"></a>00119 
<a name="l00120"></a>00120   <span class="comment">// return parameter vector. the size should be dsize();</span>
<a name="l00121"></a>00121   <span class="keyword">virtual</span> <span class="keyword">const</span> <span class="keywordtype">float</span> *<a class="code" href="classCRFPP_1_1Tagger.html#b87c8c13cb0efcfaa7b6a6d1f1b75f33">weight_vector</a>() <span class="keyword">const</span> = 0;
<a name="l00122"></a>00122 <span class="preprocessor">#endif</span>
<a name="l00123"></a>00123 <span class="preprocessor"></span>
<a name="l00124"></a>00124   <span class="comment">// set vlevel</span>
<a name="l00125"></a>00125   <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classCRFPP_1_1Tagger.html#f729184ba777adec3cc48c717739e79a">set_vlevel</a>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> <a class="code" href="classCRFPP_1_1Tagger.html#2c7e75615ca8d666da146be1b3570098">vlevel</a>) = 0;
<a name="l00126"></a>00126 
<a name="l00127"></a>00127   <span class="comment">// get vlevel</span>
<a name="l00128"></a>00128   <span class="keyword">virtual</span> <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> <a class="code" href="classCRFPP_1_1Tagger.html#2c7e75615ca8d666da146be1b3570098">vlevel</a>() <span class="keyword">const</span> = 0;
<a name="l00129"></a>00129 
<a name="l00130"></a>00130   <span class="comment">// set cost factor</span>
<a name="l00131"></a>00131   <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classCRFPP_1_1Tagger.html#d2d6c8e09a5215de7b428633d7ffa7fa">set_cost_factor</a>(<span class="keywordtype">float</span> <a class="code" href="classCRFPP_1_1Tagger.html#4ac9f6bbfad372e4594960c7c7ce568c">cost_factor</a>) = 0;
<a name="l00132"></a>00132 
<a name="l00133"></a>00133   <span class="comment">// get cost factor</span>
<a name="l00134"></a>00134   <span class="keyword">virtual</span> <span class="keywordtype">float</span> <a class="code" href="classCRFPP_1_1Tagger.html#4ac9f6bbfad372e4594960c7c7ce568c">cost_factor</a>() <span class="keyword">const</span> = 0;
<a name="l00135"></a>00135 
<a name="l00136"></a>00136   <span class="comment">// set nbest</span>
<a name="l00137"></a>00137   <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classCRFPP_1_1Tagger.html#83b6bda1a107819b560047ba22216d78">set_nbest</a>(<span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#7b5de55779ac461bea10a250e2c932cc">nbest</a>) = 0;
<a name="l00138"></a>00138 
<a name="l00139"></a>00139   <span class="comment">// get nbest</span>
<a name="l00140"></a>00140   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#7b5de55779ac461bea10a250e2c932cc">nbest</a>() <span class="keyword">const</span> = 0;
<a name="l00141"></a>00141 
<a name="l00142"></a>00142   <span class="comment">// add one line to the current context</span>
<a name="l00143"></a>00143   <span class="keyword">virtual</span> <span class="keywordtype">bool</span> <span class="keyword">add</span>(<span class="keyword">const</span> <span class="keywordtype">char</span>* str) = 0;
<a name="l00144"></a>00144 
<a name="l00145"></a>00145   <span class="comment">// return size of tokens(lines)</span>
<a name="l00146"></a>00146   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#0bb307faa0e68f71ca1bb6a992bc01e4">size</a>() <span class="keyword">const</span> = 0;
<a name="l00147"></a>00147 
<a name="l00148"></a>00148   <span class="comment">// return size of column</span>
<a name="l00149"></a>00149   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#31403e849acb03bc109367f03f052ec3">xsize</a>() <span class="keyword">const</span> = 0;
<a name="l00150"></a>00150 
<a name="l00151"></a>00151   <span class="comment">// return size of features</span>
<a name="l00152"></a>00152   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#141878c42290862182f5df5023348763">dsize</a>() <span class="keyword">const</span> = 0;
<a name="l00153"></a>00153 
<a name="l00154"></a>00154   <span class="comment">// return output tag-id of i-th token</span>
<a name="l00155"></a>00155   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#64474629e38769c9b103b6680a5cebbd">result</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00156"></a>00156 
<a name="l00157"></a>00157   <span class="comment">// return answer tag-id of i-th token if it is available</span>
<a name="l00158"></a>00158   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#cb0a26699ae23f463b5d7a8d844a8941">answer</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00159"></a>00159 
<a name="l00160"></a>00160   <span class="comment">// alias of result(i)</span>
<a name="l00161"></a>00161   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#89885d85c0914160fb43fa4d80a2f73d">y</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00162"></a>00162 
<a name="l00163"></a>00163   <span class="comment">// return output tag of i-th token as string</span>
<a name="l00164"></a>00164   <span class="keyword">virtual</span> <span class="keyword">const</span> <span class="keywordtype">char</span>*   <a class="code" href="classCRFPP_1_1Tagger.html#4a08dac43a80be3ba79fd08cf52fa228">y2</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00165"></a>00165 
<a name="l00166"></a>00166   <span class="comment">// return i-th tag-id as string</span>
<a name="l00167"></a>00167   <span class="keyword">virtual</span> <span class="keyword">const</span> <span class="keywordtype">char</span>*   <a class="code" href="classCRFPP_1_1Tagger.html#92351d6e270ad6897986e8c9898240b8">yname</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00168"></a>00168 
<a name="l00169"></a>00169   <span class="comment">// return token at [i,j] as string(i:token j:column)</span>
<a name="l00170"></a>00170   <span class="keyword">virtual</span> <span class="keyword">const</span> <span class="keywordtype">char</span>*   <a class="code" href="classCRFPP_1_1Tagger.html#f10eeb55a0465e553f99fb83191fd6cc">x</a>(<span class="keywordtype">size_t</span> i, <span class="keywordtype">size_t</span> j) <span class="keyword">const</span> = 0;
<a name="l00171"></a>00171 
<a name="l00172"></a>00172 <span class="preprocessor">#ifndef SWIG</span>
<a name="l00173"></a>00173 <span class="preprocessor"></span>  <span class="comment">// return an array of strings at i-th tokens</span>
<a name="l00174"></a>00174   <span class="keyword">virtual</span> <span class="keyword">const</span> <span class="keywordtype">char</span>**  <a class="code" href="classCRFPP_1_1Tagger.html#f10eeb55a0465e553f99fb83191fd6cc">x</a>(<span class="keywordtype">size_t</span>) <span class="keyword">const</span> = 0;
<a name="l00175"></a>00175 <span class="preprocessor">#endif</span>
<a name="l00176"></a>00176 <span class="preprocessor"></span>
<a name="l00177"></a>00177   <span class="comment">// return size of output tags</span>
<a name="l00178"></a>00178   <span class="keyword">virtual</span> <span class="keywordtype">size_t</span> <a class="code" href="classCRFPP_1_1Tagger.html#00d8320a6d642291aa3c2cce5d37ed8a">ysize</a>() <span class="keyword">const</span> = 0;
<a name="l00179"></a>00179 
<a name="l00180"></a>00180   <span class="comment">// return marginal probability of j-th tag id at i-th token</span>
<a name="l00181"></a>00181   <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classCRFPP_1_1Tagger.html#4dc84e06429db05a60fa6092ee992728">prob</a>(<span class="keywordtype">size_t</span> i, <span class="keywordtype">size_t</span> j) <span class="keyword">const</span> = 0;
<a name="l00182"></a>00182 
<a name="l00183"></a>00183   <span class="comment">// return marginal probability of output tag at i-th token</span>
<a name="l00184"></a>00184   <span class="comment">// same as prob(i, tagger-&gt;y(i));</span>
<a name="l00185"></a>00185   <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classCRFPP_1_1Tagger.html#4dc84e06429db05a60fa6092ee992728">prob</a>(<span class="keywordtype">size_t</span> i) <span class="keyword">const</span> = 0;
<a name="l00186"></a>00186 
<a name="l00187"></a>00187   <span class="comment">// return conditional probability of enter output</span>
<a name="l00188"></a>00188   <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classCRFPP_1_1Tagger.html#4dc84e06429db05a60fa6092ee992728">prob</a>() <span class="keyword">const</span> = 0;
<a name="l00189"></a>00189 
<a name="l00190"></a>00190   <span class="comment">// return forward log-prob of the j-th tag at i-th token</span>
<a name="l00191"></a>00191   <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classCRFPP_1_1Tagger.html#0fd1381e54737522383e3345d0460946">alpha</a>(<span class="keywordtype">size_t</span> i, <span class="keywordtype">size_t</span> j) <span class="keyword">const</span> = 0;
<a name="l00192"></a>00192 
<a name="l00193"></a>00193   <span class="comment">// return backward log-prob of the j-th tag at i-th token</span>
<a name="l00194"></a>00194   <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classCRFPP_1_1Tagger.html#98e054a8f5fb2eeb629fa2d7067aae25">beta</a>(<span class="keywordtype">size_t</span> i, <span class="keywordtype">size_t</span> j) <span class="keyword">const</span> = 0;
<a name="l00195"></a>00195 
<a name="l00196"></a>00196   <span class="comment">// return emission cost of the j-th tag at i-th token</span>