tree__cand__gen_8h-source.html

来自「这是一个用于数据挖掘的常用算法的模板库（数据挖掘的C++模板库for UNIX)」· HTML 代码 · 共 189 行 · 第 1/2 页

<a name="l00082"></a>00082         
<a name="l00083"></a>00083         <span class="keywordflow">if</span>(!cand_pats[i]-&gt;is_valid(minsup)) {
<a name="l00084"></a>00084           <span class="keyword">delete</span> cand_pats[i];
<a name="l00085"></a>00085           cand_pats[i]=0;
<a name="l00086"></a>00086         }
<a name="l00087"></a>00087         <span class="keywordflow">else</span> {
<a name="l00088"></a>00088           pat_fam_i.push_back(cand_pats[i]);
<a name="l00089"></a>00089           <span class="keywordflow">if</span>(cand_pats[i]-&gt;is_canonical() &amp;&amp; cand_pats[i]-&gt;is_freq(minsup)) {
<a name="l00090"></a>00090             <span class="comment">// freq_pats.push_back(cand_pats[i]);</span>
<a name="l00091"></a>00091             freq_pats_count++;
<a name="l00092"></a>00092             <span class="keywordflow">if</span>(print)
<a name="l00093"></a>00093               cout &lt;&lt; cand_pats[i];
<a name="l00094"></a>00094           }
<a name="l00095"></a>00095         }
<a name="l00096"></a>00096       }
<a name="l00097"></a>00097       
<a name="l00098"></a>00098       <span class="comment">// cleanup for this iteration</span>
<a name="l00099"></a>00099       <span class="keyword">delete</span>[] cand_pats;
<a name="l00100"></a>00100       
<a name="l00101"></a>00101     }<span class="comment">//end for it_j</span>
<a name="l00102"></a>00102     
<a name="l00103"></a>00103     <span class="comment">// recursive call</span>
<a name="l00104"></a>00104     <span class="keywordflow">if</span>(!pat_fam_i.empty())
<a name="l00105"></a>00105       cand_gen(pat_fam_i, pat_fam_i, minsup, freq_pats, cs);
<a name="l00106"></a>00106     
<a name="l00107"></a>00107   }<span class="comment">//end for it_i</span>
<a name="l00108"></a>00108   
<a name="l00109"></a>00109   <span class="comment">// cleanup VATs for entire equivalence class</span>
<a name="l00110"></a>00110   <span class="keywordflow">for</span>(it_i=Fk.begin(); it_i!=Fk.end(); it_i++) {
<a name="l00111"></a>00111     cs.delete_vat(*it_i);
<a name="l00112"></a>00112     TREE_PATTERN* pat_i=*it_i;
<a name="l00113"></a>00113     <span class="keywordflow">if</span>(!pat_i-&gt;is_freq(minsup) || !pat_i-&gt;is_canonical())
<a name="l00114"></a>00114       <span class="keyword">delete</span> pat_i;
<a name="l00115"></a>00115   }
<a name="l00116"></a>00116 }<span class="comment">//end cand_gen()</span>
<a name="l00117"></a>00117 
<a name="l00118"></a>00118 <span class="keyword">template</span>&lt;<span class="keyword">class</span> PP, <span class="keyword">class</span> MP, <span class="keyword">class</span> PAT_ST, <span class="keyword">template</span>&lt;<span class="keyword">typename</span>, <span class="keyword">typename</span>, <span class="keyword">typename</span>, <span class="keyword">template</span> &lt;<span class="keyword">typename</span>&gt; <span class="keyword">class </span>&gt; <span class="keyword">class </span>CC, template &lt;typename&gt; class ALLOC&gt;
<a name="l00119"></a>00119 TREE_PATTERN** join(<span class="keyword">const</span> TREE_PATTERN* pat_i, TREE_PATTERN* pat_j, <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; pos_i, <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; pos_j) {
<a name="l00120"></a>00120   <span class="comment">// NOTE: follows scheme of TreeMinerV, perhaps SLEUTH may be more elegant ??</span>
<a name="l00121"></a>00121   
<a name="l00122"></a>00122   <span class="comment">// NOTE: convention is that cand_pats[0] is cousin extension, </span>
<a name="l00123"></a>00123   <span class="comment">// cand_pats[1] is child extension</span>
<a name="l00124"></a>00124   <span class="comment">// This convention is used by vat::intersect</span>
<a name="l00125"></a>00125   <span class="keyword">const</span> <span class="keyword">typename</span> TREE_PATTERN::EDGE_T E_LBL=0; <span class="comment">//default, defunt edge label for trees</span>
<a name="l00126"></a>00126   TREE_PATTERN** cand_pats=<span class="keyword">new</span> TREE_PATTERN*[2];
<a name="l00127"></a>00127   
<a name="l00128"></a>00128   <span class="keywordflow">if</span>(pos_i==pos_j &amp;&amp; pat_i-&gt;size()&gt;1) {
<a name="l00129"></a>00129     <span class="comment">// case Ia of paper</span>
<a name="l00130"></a>00130     <span class="comment">// construct first candidate - cousin extension</span>
<a name="l00131"></a>00131     cand_pats[0]=pat_i-&gt;clone();
<a name="l00132"></a>00132     <span class="keyword">const</span> <span class="keyword">typename</span> TREE_PATTERN::VERTEX_T&amp; new_v=pat_j-&gt;rmost_vertex();
<a name="l00133"></a>00133     <span class="keywordtype">int</span> rvid=cand_pats[0]-&gt;add_vertex(new_v);
<a name="l00134"></a>00134     cand_pats[0]-&gt;add_out_edge(pos_j, rvid, E_LBL);
<a name="l00135"></a>00135     cand_pats[0]-&gt;add_in_edge(rvid, pos_j, E_LBL);
<a name="l00136"></a>00136     
<a name="l00137"></a>00137     <span class="comment">//construct second candidate - child extension</span>
<a name="l00138"></a>00138     cand_pats[1]=pat_i-&gt;clone();
<a name="l00139"></a>00139     rvid=cand_pats[1]-&gt;add_vertex(new_v);
<a name="l00140"></a>00140     cand_pats[1]-&gt;add_out_edge(pat_i-&gt;rmost_vid(), rvid, E_LBL);
<a name="l00141"></a>00141     cand_pats[1]-&gt;add_in_edge(rvid, pat_i-&gt;rmost_vid(), E_LBL);
<a name="l00142"></a>00142     
<a name="l00143"></a>00143     <span class="keywordflow">return</span> cand_pats;
<a name="l00144"></a>00144   }<span class="comment">//end if case Ia</span>
<a name="l00145"></a>00145   
<a name="l00146"></a>00146   <span class="comment">// execution reaches this point iff num_cands=1</span>
<a name="l00147"></a>00147   <span class="keywordflow">if</span>(pos_i&gt;pos_j) {
<a name="l00148"></a>00148     <span class="comment">// case II - cousin extension</span>
<a name="l00149"></a>00149     <span class="comment">// construct candidate</span>
<a name="l00150"></a>00150     cand_pats[0]=pat_i-&gt;clone();
<a name="l00151"></a>00151     <span class="keyword">const</span> <span class="keyword">typename</span> TREE_PATTERN::VERTEX_T&amp; new_v=pat_j-&gt;rmost_vertex();
<a name="l00152"></a>00152     <span class="keywordtype">int</span> rvid=cand_pats[0]-&gt;add_vertex(new_v);
<a name="l00153"></a>00153     cand_pats[0]-&gt;add_out_edge(pos_j, rvid, E_LBL);
<a name="l00154"></a>00154     cand_pats[0]-&gt;add_in_edge(rvid, pos_j, E_LBL);
<a name="l00155"></a>00155     cand_pats[1]=0;
<a name="l00156"></a>00156     <span class="keywordflow">return</span> cand_pats;
<a name="l00157"></a>00157   }
<a name="l00158"></a>00158   
<a name="l00159"></a>00159   <span class="comment">// assert: this is case Ib - child extension</span>
<a name="l00160"></a>00160   cand_pats[0]=0;
<a name="l00161"></a>00161   cand_pats[1]=pat_i-&gt;clone();
<a name="l00162"></a>00162   <span class="keyword">const</span> <span class="keyword">typename</span> TREE_PATTERN::VERTEX_T&amp; new_v=pat_j-&gt;rmost_vertex();
<a name="l00163"></a>00163   <span class="keywordtype">int</span> rvid=cand_pats[1]-&gt;add_vertex(new_v);
<a name="l00164"></a>00164   cand_pats[1]-&gt;add_out_edge(pat_i-&gt;rmost_vid(), rvid, E_LBL);
<a name="l00165"></a>00165   cand_pats[1]-&gt;add_in_edge(rvid, pat_i-&gt;rmost_vid(), E_LBL);
<a name="l00166"></a>00166   <span class="keywordflow">return</span> cand_pats;
<a name="l00167"></a>00167 }<span class="comment">//end join()</span>
<a name="l00168"></a>00168 
<a name="l00169"></a>00169 <span class="preprocessor">#endif</span>
</pre></div><hr size="1"><address style="align: right;"><small>Generated on Wed Jul 26 14:01:08 2006 for DMTL by&nbsp;
<a href="http://www.doxygen.org/index.html">
<img src="doxygen.png" alt="doxygen" align="middle" border="0"></a> 1.4.7 </small></address>
</body>
</html>