| <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
| <html xmlns="http://www.w3.org/1999/xhtml"> |
| <head> |
| <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> |
| <meta http-equiv="X-UA-Compatible" content="IE=9"/> |
| <meta name="generator" content="Doxygen 1.8.4"/> |
| <title>MADlib: plda.sql_in Source File</title> |
| <link href="tabs.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="jquery.js"></script> |
| <script type="text/javascript" src="dynsections.js"></script> |
| <link href="navtree.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="resize.js"></script> |
| <script type="text/javascript" src="navtree.js"></script> |
| <script type="text/javascript"> |
| $(document).ready(initResizable); |
| $(window).load(resizeHeight); |
| </script> |
| <link href="search/search.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript" src="search/search.js"></script> |
| <script type="text/javascript"> |
| $(document).ready(function() { searchBox.OnSelectItem(0); }); |
| </script> |
| <script type="text/x-mathjax-config"> |
| MathJax.Hub.Config({ |
| extensions: ["tex2jax.js", "TeX/AMSmath.js", "TeX/AMSsymbols.js"], |
| jax: ["input/TeX","output/HTML-CSS"], |
| }); |
| </script><script src="../mathjax/MathJax.js"></script> |
| <link href="doxygen.css" rel="stylesheet" type="text/css" /> |
| </head> |
| <body> |
| <div id="top"><!-- do not remove this div, it is closed by doxygen! --> |
| <div id="titlearea"> |
| <table cellspacing="0" cellpadding="0"> |
| <tbody> |
| <tr style="height: 56px;"> |
| <td style="padding-left: 0.5em;"> |
| <div id="projectname">MADlib |
|  <span id="projectnumber">0.5</span> |
| </div> |
| <div id="projectbrief">User Documentation</div> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <!-- end header part --> |
| <!-- Generated by Doxygen 1.8.4 --> |
| <script type="text/javascript"> |
| var searchBox = new SearchBox("searchBox", "search",false,'Search'); |
| </script> |
| <div id="navrow1" class="tabs"> |
| <ul class="tablist"> |
| <li><a href="index.html"><span>Main Page</span></a></li> |
| <li><a href="modules.html"><span>Modules</span></a></li> |
| <li> |
| <div id="MSearchBox" class="MSearchBoxInactive"> |
| <span class="left"> |
| <img id="MSearchSelect" src="search/mag_sel.png" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| alt=""/> |
| <input type="text" id="MSearchField" value="Search" accesskey="S" |
| onfocus="searchBox.OnSearchFieldFocus(true)" |
| onblur="searchBox.OnSearchFieldFocus(false)" |
| onkeyup="searchBox.OnSearchFieldChange(event)"/> |
| </span><span class="right"> |
| <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a> |
| </span> |
| </div> |
| </li> |
| </ul> |
| </div> |
| </div><!-- top --> |
| <div id="side-nav" class="ui-resizable side-nav-resizable"> |
| <div id="nav-tree"> |
| <div id="nav-tree-contents"> |
| <div id="nav-sync" class="sync"></div> |
| </div> |
| </div> |
| <div id="splitbar" style="-moz-user-select:none;" |
| class="ui-resizable-handle"> |
| </div> |
| </div> |
| <script type="text/javascript"> |
| $(document).ready(function(){initNavTree('plda_8sql__in_source.html','');}); |
| </script> |
| <div id="doc-content"> |
| <!-- window showing the filter options --> |
| <div id="MSearchSelectWindow" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| onkeydown="return searchBox.OnSearchSelectKey(event)"> |
| <a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(0)"><span class="SelectionMark"> </span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark"> </span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark"> </span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark"> </span>Groups</a></div> |
| |
| <!-- iframe showing the search results (closed by default) --> |
| <div id="MSearchResultsWindow"> |
| <iframe src="javascript:void(0)" frameborder="0" |
| name="MSearchResults" id="MSearchResults"> |
| </iframe> |
| </div> |
| |
| <div class="header"> |
| <div class="headertitle"> |
| <div class="title">plda.sql_in</div> </div> |
| </div><!--header--> |
| <div class="contents"> |
| <a href="plda_8sql__in.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">/* ----------------------------------------------------------------------- */</span><span class="comment">/** </span></div> |
| <div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment"> * @file plda.sql_in</span></div> |
| <div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00005"></a><span class="lineno"> 5</span> <span class="comment"> * @brief SQL functions for parallel Latent Dirichlet Allocation</span></div> |
| <div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="comment"> * @date April 2011</span></div> |
| <div class="line"><a name="l00007"></a><span class="lineno"> 7</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="comment"> * @sa For an introduction to Latent Dirichlet Allocation models, see the </span></div> |
| <div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="comment"> module description \ref grp_plda.</span></div> |
| <div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00011"></a><span class="lineno"> 11</span> <span class="comment"> */</span><span class="comment">/* ------------------------------------------------------------------------*/</span></div> |
| <div class="line"><a name="l00012"></a><span class="lineno"> 12</span> </div> |
| <div class="line"><a name="l00013"></a><span class="lineno"> 13</span> m4_include(`SQLCommon.m4<span class="stringliteral">')</span></div> |
| <div class="line"><a name="l00014"></a><span class="lineno"> 14</span> <span class="stringliteral"></span><span class="comment"></span></div> |
| <div class="line"><a name="l00015"></a><span class="lineno"> 15</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00017"></a><span class="lineno"> 17</span> <span class="comment">@addtogroup grp_plda</span></div> |
| <div class="line"><a name="l00018"></a><span class="lineno"> 18</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00019"></a><span class="lineno"> 19</span> <span class="comment">@about</span></div> |
| <div class="line"><a name="l00020"></a><span class="lineno"> 20</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00021"></a><span class="lineno"> 21</span> <span class="comment">Latent Dirichlet Allocation (LDA) is an interesting generative probabilistic </span></div> |
| <div class="line"><a name="l00022"></a><span class="lineno"> 22</span> <span class="comment">model for natural texts that has received a lot of attention in recent years. </span></div> |
| <div class="line"><a name="l00023"></a><span class="lineno"> 23</span> <span class="comment">The model is quite versatile, having found uses in problems like automated </span></div> |
| <div class="line"><a name="l00024"></a><span class="lineno"> 24</span> <span class="comment">topic discovery, collaborative filtering, and document classification.</span></div> |
| <div class="line"><a name="l00025"></a><span class="lineno"> 25</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00026"></a><span class="lineno"> 26</span> <span class="comment">The LDA model posits that each document is associated with a mixture of various </span></div> |
| <div class="line"><a name="l00027"></a><span class="lineno"> 27</span> <span class="comment">topics (e.g. a document is related to Topic 1 with probability 0.7, and Topic 2 with </span></div> |
| <div class="line"><a name="l00028"></a><span class="lineno"> 28</span> <span class="comment">probability 0.3), and that each word in the document is attributable to one </span></div> |
| <div class="line"><a name="l00029"></a><span class="lineno"> 29</span> <span class="comment">of the document's topics. There is a (symmetric) Dirichlet prior with parameter </span></div> |
| <div class="line"><a name="l00030"></a><span class="lineno"> 30</span> <span class="comment">\f$ \alpha \f$ on each document's topic mixture. In addition, there is another </span></div> |
| <div class="line"><a name="l00031"></a><span class="lineno"> 31</span> <span class="comment">(symmateric) Dirichlet prior with parameter \f$ \eta \f$ on the distribution </span></div> |
| <div class="line"><a name="l00032"></a><span class="lineno"> 32</span> <span class="comment">of words for each topic. The following generative process then defines a distribution </span></div> |
| <div class="line"><a name="l00033"></a><span class="lineno"> 33</span> <span class="comment">over a corpus of documents. First sample, for each topic \f$ i \f$, a per-topic word distribution </span></div> |
| <div class="line"><a name="l00034"></a><span class="lineno"> 34</span> <span class="comment">\f$ \Phi_i \f$ from the Dirichlet(\f$\eta\f$) prior. </span></div> |
| <div class="line"><a name="l00035"></a><span class="lineno"> 35</span> <span class="comment">Then for each document:</span></div> |
| <div class="line"><a name="l00036"></a><span class="lineno"> 36</span> <span class="comment">-# Sample a document length N from a suitable distribution, say, Poisson.</span></div> |
| <div class="line"><a name="l00037"></a><span class="lineno"> 37</span> <span class="comment">-# Sample a topic mixture \f$ \theta \f$ for the document from the Dirichlet(\f$\alpha\f$) distribution.</span></div> |
| <div class="line"><a name="l00038"></a><span class="lineno"> 38</span> <span class="comment">-# For each of the N words:</span></div> |
| <div class="line"><a name="l00039"></a><span class="lineno"> 39</span> <span class="comment"> -# Sample a topic \f$ z_n \f$ from the multinomial topic distribution \f$ \theta \f$.</span></div> |
| <div class="line"><a name="l00040"></a><span class="lineno"> 40</span> <span class="comment"> -# Sample a word \f$ w_n \f$ from the multinomial word distribution \f$ \Phi_{z_n} \f$ associated with topic \f$ z_n \f$.</span></div> |
| <div class="line"><a name="l00041"></a><span class="lineno"> 41</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00042"></a><span class="lineno"> 42</span> <span class="comment">In practice, only the words in each document are observable. The topic mixture of </span></div> |
| <div class="line"><a name="l00043"></a><span class="lineno"> 43</span> <span class="comment">each document and the topic for each word in each document are latent unobservable </span></div> |
| <div class="line"><a name="l00044"></a><span class="lineno"> 44</span> <span class="comment">variables that need to be inferred from the observables, and this is the problem</span></div> |
| <div class="line"><a name="l00045"></a><span class="lineno"> 45</span> <span class="comment">people refer to when they talk about the inference problem for LDA. Exact inference</span></div> |
| <div class="line"><a name="l00046"></a><span class="lineno"> 46</span> <span class="comment">is intractable, but several approximate inference algorithms for LDA have been</span></div> |
| <div class="line"><a name="l00047"></a><span class="lineno"> 47</span> <span class="comment">developed. The simple and effective Gibbs sampling algorithm described in </span></div> |
| <div class="line"><a name="l00048"></a><span class="lineno"> 48</span> <span class="comment">Griffiths and Steyvers [2] appears to be the current algorithm of choice. Our </span></div> |
| <div class="line"><a name="l00049"></a><span class="lineno"> 49</span> <span class="comment">parallel implementation of LDA comes from Wang et al [3], which is essentially</span></div> |
| <div class="line"><a name="l00050"></a><span class="lineno"> 50</span> <span class="comment">a straightforward parallelisation of the Gibbs sampling algorithm.</span></div> |
| <div class="line"><a name="l00051"></a><span class="lineno"> 51</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00052"></a><span class="lineno"> 52</span> <span class="comment">See also http://code.google.com/p/plda/.</span></div> |
| <div class="line"><a name="l00053"></a><span class="lineno"> 53</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00054"></a><span class="lineno"> 54</span> <span class="comment">@input</span></div> |
| <div class="line"><a name="l00055"></a><span class="lineno"> 55</span> <span class="comment">The \b corpus to be analyzed is expected to be of the following form:</span></div> |
| <div class="line"><a name="l00056"></a><span class="lineno"> 56</span> <span class="comment"><pre>{TABLE|VIEW} <em>datatable</em> (</span></div> |
| <div class="line"><a name="l00057"></a><span class="lineno"> 57</span> <span class="comment"> <em>id</em> INTEGER,</span></div> |
| <div class="line"><a name="l00058"></a><span class="lineno"> 58</span> <span class="comment"> <em>contents</em> INTEGER[],</span></div> |
| <div class="line"><a name="l00059"></a><span class="lineno"> 59</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00060"></a><span class="lineno"> 60</span> <span class="comment">)</pre></span></div> |
| <div class="line"><a name="l00061"></a><span class="lineno"> 61</span> <span class="comment">where \c id refers to the document ID, and \c contents is an integer array that specifies the words in the document using the index from</span></div> |
| <div class="line"><a name="l00062"></a><span class="lineno"> 62</span> <span class="comment">the dictionary. Words must be represented using positive numbers.</span></div> |
| <div class="line"><a name="l00063"></a><span class="lineno"> 63</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00064"></a><span class="lineno"> 64</span> <span class="comment">The \b dictionary that indexes all the words found in the corpus is of the following form:</span></div> |
| <div class="line"><a name="l00065"></a><span class="lineno"> 65</span> <span class="comment"><pre>{TABLE|VIEW} <em>dicttable</em> (</span></div> |
| <div class="line"><a name="l00066"></a><span class="lineno"> 66</span> <span class="comment"> <em>dict</em> TEXT[],</span></div> |
| <div class="line"><a name="l00067"></a><span class="lineno"> 67</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00068"></a><span class="lineno"> 68</span> <span class="comment">)</pre></span></div> |
| <div class="line"><a name="l00069"></a><span class="lineno"> 69</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00070"></a><span class="lineno"> 70</span> <span class="comment">@usage</span></div> |
| <div class="line"><a name="l00071"></a><span class="lineno"> 71</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00072"></a><span class="lineno"> 72</span> <span class="comment">- Topic inference is achieved through the following UDF</span></div> |
| <div class="line"><a name="l00073"></a><span class="lineno"> 73</span> <span class="comment"> <pre></span></div> |
| <div class="line"><a name="l00074"></a><span class="lineno"> 74</span> <span class="comment"> SELECT \ref plda_run('<em>datatable</em>', '<em>dicttable</em>', '<em>modeltable</em>', '<em>outputdatatable</em>', </span></div> |
| <div class="line"><a name="l00075"></a><span class="lineno"> 75</span> <span class="comment"> <em>numiter</em>, <em>numtopics</em>, <em>alpha</em>, <em>eta</em>);</span></div> |
| <div class="line"><a name="l00076"></a><span class="lineno"> 76</span> <span class="comment"> </pre></span></div> |
| <div class="line"><a name="l00077"></a><span class="lineno"> 77</span> <span class="comment"> This function stores the resulting model in <tt><em>outputdatatable</em></tt>.</span></div> |
| <div class="line"><a name="l00078"></a><span class="lineno"> 78</span> <span class="comment">- Labelling of test documents using a learned LDA model is achieved using the following UDF</span></div> |
| <div class="line"><a name="l00079"></a><span class="lineno"> 79</span> <span class="comment"> <pre></span></div> |
| <div class="line"><a name="l00080"></a><span class="lineno"> 80</span> <span class="comment"> SELECT \ref plda_label_test_documents('<em>testtable</em>', '<em>outputtable</em>', '<em>modeltable</em>', '<em>dicttable</em>',</span></div> |
| <div class="line"><a name="l00081"></a><span class="lineno"> 81</span> <span class="comment"> <em>numtopics</em>, <em>alpha</em>, <em>eta</em>);</span></div> |
| <div class="line"><a name="l00082"></a><span class="lineno"> 82</span> <span class="comment"> </pre></span></div> |
| <div class="line"><a name="l00083"></a><span class="lineno"> 83</span> <span class="comment"> This creates the following table with the assigned topic for each word in the test corpus.</span></div> |
| <div class="line"><a name="l00084"></a><span class="lineno"> 84</span> <span class="comment"> <pre></span></div> |
| <div class="line"><a name="l00085"></a><span class="lineno"> 85</span> <span class="comment"> id | contents | topics </span></div> |
| <div class="line"><a name="l00086"></a><span class="lineno"> 86</span> <span class="comment"> ----+-------------------------------+-----------------------------------</span></div> |
| <div class="line"><a name="l00087"></a><span class="lineno"> 87</span> <span class="comment"> ...</span></div> |
| <div class="line"><a name="l00088"></a><span class="lineno"> 88</span> <span class="comment"> </pre></span></div> |
| <div class="line"><a name="l00089"></a><span class="lineno"> 89</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00090"></a><span class="lineno"> 90</span> <span class="comment">@implementation</span></div> |
| <div class="line"><a name="l00091"></a><span class="lineno"> 91</span> <span class="comment">The input format for the Parallel LDA module is different from that used by the </span></div> |
| <div class="line"><a name="l00092"></a><span class="lineno"> 92</span> <span class="comment">`lda' package for R. In the `lda' package, each document is represented by two</span></div> |
| <div class="line"><a name="l00093"></a><span class="lineno"> 93</span> <span class="comment">equal dimensional integer arrays. The first array represents the words that occur</span></div> |
| <div class="line"><a name="l00094"></a><span class="lineno"> 94</span> <span class="comment">in the document, and the second array captures the number of times each word in</span></div> |
| <div class="line"><a name="l00095"></a><span class="lineno"> 95</span> <span class="comment">the first array occurs in the document.</span></div> |
| <div class="line"><a name="l00096"></a><span class="lineno"> 96</span> <span class="comment">This representation appears to have a major weakness in that all the occurrences</span></div> |
| <div class="line"><a name="l00097"></a><span class="lineno"> 97</span> <span class="comment">of a word in a document must be assigned the same topic, which is clearly not</span></div> |
| <div class="line"><a name="l00098"></a><span class="lineno"> 98</span> <span class="comment">satisfactory. Further, at the time of writing, the main learning function in the </span></div> |
| <div class="line"><a name="l00099"></a><span class="lineno"> 99</span> <span class="comment">`lda' package actually does not work correctly when the occurrence counts for words </span></div> |
| <div class="line"><a name="l00100"></a><span class="lineno"> 100</span> <span class="comment">aren't one.</span></div> |
| <div class="line"><a name="l00101"></a><span class="lineno"> 101</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00102"></a><span class="lineno"> 102</span> <span class="comment">There is a script called generateTestCases.cc that can be used to generate some</span></div> |
| <div class="line"><a name="l00103"></a><span class="lineno"> 103</span> <span class="comment">simple test documents to validate the correctness and effiency of the parallel</span></div> |
| <div class="line"><a name="l00104"></a><span class="lineno"> 104</span> <span class="comment">LDA implementation.</span></div> |
| <div class="line"><a name="l00105"></a><span class="lineno"> 105</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00106"></a><span class="lineno"> 106</span> <span class="comment">@examp</span></div> |
| <div class="line"><a name="l00107"></a><span class="lineno"> 107</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00108"></a><span class="lineno"> 108</span> <span class="comment">We now give a usage example.</span></div> |
| <div class="line"><a name="l00109"></a><span class="lineno"> 109</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00110"></a><span class="lineno"> 110</span> <span class="comment">-# As a first step, we need to prepare a corpus and dictionary in the appropriate structure.</span></div> |
| <div class="line"><a name="l00111"></a><span class="lineno"> 111</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00112"></a><span class="lineno"> 112</span> <span class="comment">sql> CREATE TABLE MADLIB_SCHEMA.plda_mydict ( dict text[] ) DISTRIBUTED RANDOMLY;</span></div> |
| <div class="line"><a name="l00113"></a><span class="lineno"> 113</span> <span class="comment">sql> INSERT into MADLIB_SCHEMA.plda_mydict values</span></div> |
| <div class="line"><a name="l00114"></a><span class="lineno"> 114</span> <span class="comment"> ('{human,machine,interface,for,abc,computer,applications,a,survey,of,</span></div> |
| <div class="line"><a name="l00115"></a><span class="lineno"> 115</span> <span class="comment"> user,opinion,system,response,time,the,eps,management,and,engineering,</span></div> |
| <div class="line"><a name="l00116"></a><span class="lineno"> 116</span> <span class="comment"> testing,relation,perceived,to,error,generation,random,binary,order,tree,</span></div> |
| <div class="line"><a name="l00117"></a><span class="lineno"> 117</span> <span class="comment"> intersection,graph,path,in,minor,IV,widths,well,quasi}');</span></div> |
| <div class="line"><a name="l00118"></a><span class="lineno"> 118</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00119"></a><span class="lineno"> 119</span> <span class="comment">sql> CREATE TABLE MADLIB_SCHEMA.plda_mycorpus ( id int4, contents int4[] );</span></div> |
| <div class="line"><a name="l00120"></a><span class="lineno"> 120</span> <span class="comment">sql> INSERT INTO MADLIB_SCHEMA.plda_mycorpus VALUES </span></div> |
| <div class="line"><a name="l00121"></a><span class="lineno"> 121</span> <span class="comment"> (1, '{1,2,3,4,5,6,7}'), </span></div> |
| <div class="line"><a name="l00122"></a><span class="lineno"> 122</span> <span class="comment"> (2, '{8,9,10,11,12,10,6,13,14,15}'),</span></div> |
| <div class="line"><a name="l00123"></a><span class="lineno"> 123</span> <span class="comment"> (3, '{16,17,11,3,18,13}'),</span></div> |
| <div class="line"><a name="l00124"></a><span class="lineno"> 124</span> <span class="comment"> (4, '{13,19,1,13,20,21,10,17}'),</span></div> |
| <div class="line"><a name="l00125"></a><span class="lineno"> 125</span> <span class="comment"> (5, '{22,10,11,23,14,15,24,25,18}'),</span></div> |
| <div class="line"><a name="l00126"></a><span class="lineno"> 126</span> <span class="comment"> (6, '{16,26,10,27,28,29,30}'),</span></div> |
| <div class="line"><a name="l00127"></a><span class="lineno"> 127</span> <span class="comment"> (7, '{16,31,32,10,33,34,30}'),</span></div> |
| <div class="line"><a name="l00128"></a><span class="lineno"> 128</span> <span class="comment"> (8, '{32,35,36,37,10,30,19,38,39,29}'),</span></div> |
| <div class="line"><a name="l00129"></a><span class="lineno"> 129</span> <span class="comment"> (9, '{32,35,8,9}') ;</span></div> |
| <div class="line"><a name="l00130"></a><span class="lineno"> 130</span> <span class="comment"> \endcode </span></div> |
| <div class="line"><a name="l00131"></a><span class="lineno"> 131</span> <span class="comment">-# To perform inference, we call the plda_run() function with the appropriate parameters. </span></div> |
| <div class="line"><a name="l00132"></a><span class="lineno"> 132</span> <span class="comment"> Here is an example.</span></div> |
| <div class="line"><a name="l00133"></a><span class="lineno"> 133</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00134"></a><span class="lineno"> 134</span> <span class="comment">sql> select MADLIB_SCHEMA.plda_run('MADLIB_SCHEMA.plda_mycorpus', 'MADLIB_SCHEMA.plda_mydict', </span></div> |
| <div class="line"><a name="l00135"></a><span class="lineno"> 135</span> <span class="comment"> 'MADLIB_SCHEMA.plda_mymodel', 'MADLIB_SCHEMA.plda_corpus', </span></div> |
| <div class="line"><a name="l00136"></a><span class="lineno"> 136</span> <span class="comment"> 30,10,0.5,0.5);</span></div> |
| <div class="line"><a name="l00137"></a><span class="lineno"> 137</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00138"></a><span class="lineno"> 138</span> <span class="comment"> After a successful run of the plda_run() function, the most probable words associated </span></div> |
| <div class="line"><a name="l00139"></a><span class="lineno"> 139</span> <span class="comment"> with each topic are displayed. Other results of the learning process can be obtained </span></div> |
| <div class="line"><a name="l00140"></a><span class="lineno"> 140</span> <span class="comment"> by running the following commands. Here we assume the modeltable and outputdatatable </span></div> |
| <div class="line"><a name="l00141"></a><span class="lineno"> 141</span> <span class="comment"> parameters to plda_run() are 'MADLIB_SCHEMA.plda_mymodel' and 'MADLIB_SCHEMA.plda_corpus' respectively.</span></div> |
| <div class="line"><a name="l00142"></a><span class="lineno"> 142</span> <span class="comment"> -# The topic assignments for each document can be obtained as follows:</span></div> |
| <div class="line"><a name="l00143"></a><span class="lineno"> 143</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00144"></a><span class="lineno"> 144</span> <span class="comment"> sql> select id, (topics).topics from MADLIB_SCHEMA.plda_corpus;</span></div> |
| <div class="line"><a name="l00145"></a><span class="lineno"> 145</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00146"></a><span class="lineno"> 146</span> <span class="comment"> -# The topic distribution of each document can be obtained as follows:</span></div> |
| <div class="line"><a name="l00147"></a><span class="lineno"> 147</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00148"></a><span class="lineno"> 148</span> <span class="comment"> sql> select id, (topics).topic_d from MADLIB_SCHEMA.plda_corpus;</span></div> |
| <div class="line"><a name="l00149"></a><span class="lineno"> 149</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00150"></a><span class="lineno"> 150</span> <span class="comment"> -# The number of times each word was assigned to a given topic in the whole corpus can </span></div> |
| <div class="line"><a name="l00151"></a><span class="lineno"> 151</span> <span class="comment"> be computed as follows:</span></div> |
| <div class="line"><a name="l00152"></a><span class="lineno"> 152</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00153"></a><span class="lineno"> 153</span> <span class="comment"> sql> select ss.i, MADLIB_SCHEMA.plda_word_topic_distrn(gcounts,$numtopics,ss.i) </span></div> |
| <div class="line"><a name="l00154"></a><span class="lineno"> 154</span> <span class="comment"> from MADLIB_SCHEMA.plda_mymodel, </span></div> |
| <div class="line"><a name="l00155"></a><span class="lineno"> 155</span> <span class="comment"> (select generate_series(1,$dictsize) i) as ss;</span></div> |
| <div class="line"><a name="l00156"></a><span class="lineno"> 156</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00157"></a><span class="lineno"> 157</span> <span class="comment"> where $numtopics is the number of topics used in the learning process, and </span></div> |
| <div class="line"><a name="l00158"></a><span class="lineno"> 158</span> <span class="comment"> $dictsize is the size of the dictionary.</span></div> |
| <div class="line"><a name="l00159"></a><span class="lineno"> 159</span> <span class="comment"> -# The total number of words assigned to each topic in the whole corpus can be computed </span></div> |
| <div class="line"><a name="l00160"></a><span class="lineno"> 160</span> <span class="comment"> as follows:</span></div> |
| <div class="line"><a name="l00161"></a><span class="lineno"> 161</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00162"></a><span class="lineno"> 162</span> <span class="comment"> sql> select sum((topics).topic_d) topic_sums from MADLIB_SCHEMA.plda_corpus;</span></div> |
| <div class="line"><a name="l00163"></a><span class="lineno"> 163</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00164"></a><span class="lineno"> 164</span> <span class="comment">-# To use a learned LDA model to label new documents, we can use the following commands:</span></div> |
| <div class="line"><a name="l00165"></a><span class="lineno"> 165</span> <span class="comment"> \code</span></div> |
| <div class="line"><a name="l00166"></a><span class="lineno"> 166</span> <span class="comment">sql> select MADLIB_SCHEMA.plda_label_test_documents('MADLIB_SCHEMA.plda_mycorpus', 'MADLIB_SCHEMA.plda_testresult', 'MADLIB_SCHEMA.plda_mymodel', 'MADLIB_SCHEMA.plda_mydict', 10,0.5,0.5);</span></div> |
| <div class="line"><a name="l00167"></a><span class="lineno"> 167</span> <span class="comment">sql> select * from MADLIB_SCHEMA.plda_testresult;</span></div> |
| <div class="line"><a name="l00168"></a><span class="lineno"> 168</span> <span class="comment"> \endcode</span></div> |
| <div class="line"><a name="l00169"></a><span class="lineno"> 169</span> <span class="comment"> </span></div> |
| <div class="line"><a name="l00170"></a><span class="lineno"> 170</span> <span class="comment">@literature</span></div> |
| <div class="line"><a name="l00171"></a><span class="lineno"> 171</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00172"></a><span class="lineno"> 172</span> <span class="comment">[1] D.M. Blei, A.Y. Ng, M.I. Jordan, <em>Latent Dirichlet Allocation</em>, </span></div> |
| <div class="line"><a name="l00173"></a><span class="lineno"> 173</span> <span class="comment"> Journal of Machine Learning Research, vol. 3, pp. 993-1022, 2003.</span></div> |
| <div class="line"><a name="l00174"></a><span class="lineno"> 174</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00175"></a><span class="lineno"> 175</span> <span class="comment">[2] T. Griffiths and M. Steyvers, <em>Finding scientific topics</em>, </span></div> |
| <div class="line"><a name="l00176"></a><span class="lineno"> 176</span> <span class="comment"> PNAS, vol. 101, pp. 5228-5235, 2004.</span></div> |
| <div class="line"><a name="l00177"></a><span class="lineno"> 177</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00178"></a><span class="lineno"> 178</span> <span class="comment">[3] Y. Wang, H. Bai, M. Stanton, W-Y. Chen, and E.Y. Chang, <em>PLDA: </span></div> |
| <div class="line"><a name="l00179"></a><span class="lineno"> 179</span> <span class="comment"> Parallel Dirichlet Allocation for Large-scale Applications</em>, AAIM, 2009.</span></div> |
| <div class="line"><a name="l00180"></a><span class="lineno"> 180</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00181"></a><span class="lineno"> 181</span> <span class="comment">[4] http://en.wikipedia.org/wiki/Latent_Dirichlet_allocation</span></div> |
| <div class="line"><a name="l00182"></a><span class="lineno"> 182</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00183"></a><span class="lineno"> 183</span> <span class="comment">[5] J. Chang, Collapsed Gibbs sampling methods for topic models, R manual, 2010.</span></div> |
| <div class="line"><a name="l00184"></a><span class="lineno"> 184</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00185"></a><span class="lineno"> 185</span> <span class="comment">@sa File plda.sql_in documenting the SQL functions.</span></div> |
| <div class="line"><a name="l00186"></a><span class="lineno"> 186</span> <span class="comment"></span></div> |
| <div class="line"><a name="l00187"></a><span class="lineno"> 187</span> <span class="comment">*/</span></div> |
| <div class="line"><a name="l00188"></a><span class="lineno"> 188</span> </div> |
| <div class="line"><a name="l00189"></a><span class="lineno"> 189</span> </div> |
| <div class="line"><a name="l00190"></a><span class="lineno"> 190</span> -- The plda_topics_t data type store the assignment of topics to each word in a document,</div> |
| <div class="line"><a name="l00191"></a><span class="lineno"> 191</span> -- plus the distribution of those topics in the document.</div> |
| <div class="line"><a name="l00192"></a><span class="lineno"> 192</span> CREATE TYPE MADLIB_SCHEMA.plda_topics_t AS (</div> |
| <div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  topics int4[],</div> |
| <div class="line"><a name="l00194"></a><span class="lineno"> 194</span>  topic_d int4[]</div> |
| <div class="line"><a name="l00195"></a><span class="lineno"> 195</span> );</div> |
| <div class="line"><a name="l00196"></a><span class="lineno"> 196</span> </div> |
| <div class="line"><a name="l00197"></a><span class="lineno"> 197</span> -- Returns a zero'd array of a given dimension</div> |
| <div class="line"><a name="l00198"></a><span class="lineno"> 198</span> CREATE OR REPLACE FUNCTION MADLIB_SCHEMA.plda_zero_array(d int4) RETURNS int4[] </div> |
| <div class="line"><a name="l00199"></a><span class="lineno"> 199</span> AS <span class="stringliteral">'MODULE_PATHNAME'</span>, <span class="stringliteral">'zero_array'</span> LANGUAGE C STRICT;</div> |
| <div class="line"><a name="l00200"></a><span class="lineno"> 200</span> </div> |
| <div class="line"><a name="l00201"></a><span class="lineno"> 201</span> -- Returns the element-wise sum of two integer arrays</div> |
| <div class="line"><a name="l00202"></a><span class="lineno"> 202</span> CREATE OR REPLACE FUNCTION MADLIB_SCHEMA.plda_sum_int4array(int4[],int4[]) RETURNS int4[]</div> |
| <div class="line"><a name="l00203"></a><span class="lineno"> 203</span> AS <span class="stringliteral">'MODULE_PATHNAME'</span>, <span class="stringliteral">'sum_int4array'</span> LANGUAGE C;</div> |
| <div class="line"><a name="l00204"></a><span class="lineno"> 204</span> </div> |
| <div class="line"><a name="l00205"></a><span class="lineno"> 205</span> -- Aggregate <span class="keyword">function</span> <span class="keywordflow">for</span> computing the element-wise sum of a set of integer arrays</div> |
| <div class="line"><a name="l00206"></a><span class="lineno"> 206</span> CREATE AGGREGATE MADLIB_SCHEMA.plda_sum_int4array_agg(int4[]) </div> |
| <div class="line"><a name="l00207"></a><span class="lineno"> 207</span> (</div> |
| <div class="line"><a name="l00208"></a><span class="lineno"> 208</span>  sfunc = MADLIB_SCHEMA.plda_sum_int4array,</div> |
| <div class="line"><a name="l00209"></a><span class="lineno"> 209</span>  stype = int4[]</div> |
| <div class="line"><a name="l00210"></a><span class="lineno"> 210</span> );</div> |
| <div class="line"><a name="l00211"></a><span class="lineno"> 211</span> </div> |
| <div class="line"><a name="l00212"></a><span class="lineno"> 212</span> -- Returns an array of random topic assignments <span class="keywordflow">for</span> a given document length</div> |
| <div class="line"><a name="l00213"></a><span class="lineno"> 213</span> CREATE OR REPLACE FUNCTION MADLIB_SCHEMA.plda_random_topics(doclen int4, numtopics int4) RETURNS MADLIB_SCHEMA.plda_topics_t </div> |
| <div class="line"><a name="l00214"></a><span class="lineno"> 214</span> AS <span class="stringliteral">'MODULE_PATHNAME'</span>, <span class="stringliteral">'randomTopics'</span> LANGUAGE C STRICT;</div> |
| <div class="line"><a name="l00215"></a><span class="lineno"> 215</span> </div> |
| <div class="line"><a name="l00216"></a><span class="lineno"> 216</span> -- This <span class="keyword">function</span> assigns a randomly chosen topic to each word in a document according to </div> |
| <div class="line"><a name="l00217"></a><span class="lineno"> 217</span> -- the count statistics obtained <span class="keywordflow">for</span> the document and the whole corpus so far. </div> |
| <div class="line"><a name="l00218"></a><span class="lineno"> 218</span> -- Parameters</div> |
| <div class="line"><a name="l00219"></a><span class="lineno"> 219</span> -- doc : the document to be analysed</div> |
| <div class="line"><a name="l00220"></a><span class="lineno"> 220</span> -- topics : the topic of each word in the doc</div> |
| <div class="line"><a name="l00221"></a><span class="lineno"> 221</span> -- topic_d : the topic distribution <span class="keywordflow">for</span> the doc</div> |
| <div class="line"><a name="l00222"></a><span class="lineno"> 222</span> -- global_count : the global word-topic counts</div> |
| <div class="line"><a name="l00223"></a><span class="lineno"> 223</span> -- topic_counts : the counts of all words in the corpus in each topic</div> |
| <div class="line"><a name="l00224"></a><span class="lineno"> 224</span> -- num_topics : number of topics to be discovered</div> |
| <div class="line"><a name="l00225"></a><span class="lineno"> 225</span> -- dsize : size of dictionary</div> |
| <div class="line"><a name="l00226"></a><span class="lineno"> 226</span> -- alpha : the parameter of the Dirichlet distribution</div> |
| <div class="line"><a name="l00227"></a><span class="lineno"> 227</span> -- eta : the parameter of the Dirichlet distribution</div> |
| <div class="line"><a name="l00228"></a><span class="lineno"> 228</span> --</div> |
| <div class="line"><a name="l00229"></a><span class="lineno"> 229</span> CREATE OR REPLACE FUNCTION</div> |
| <div class="line"><a name="l00230"></a><span class="lineno"> 230</span> MADLIB_SCHEMA.plda_sample_new_topics(doc int4[], topics int4[], topic_d int4[], global_count int4[],</div> |
| <div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  topic_counts int4[], num_topics int4, dsize int4, alpha <span class="keywordtype">float</span>, eta <span class="keywordtype">float</span>) </div> |
| <div class="line"><a name="l00232"></a><span class="lineno"> 232</span> RETURNS MADLIB_SCHEMA.plda_topics_t</div> |
| <div class="line"><a name="l00233"></a><span class="lineno"> 233</span> AS <span class="stringliteral">'MODULE_PATHNAME'</span>, <span class="stringliteral">'sampleNewTopics'</span> LANGUAGE C STRICT;</div> |
| <div class="line"><a name="l00234"></a><span class="lineno"> 234</span> </div> |
| <div class="line"><a name="l00235"></a><span class="lineno"> 235</span> -- Computes the per document word-topic counts</div> |
| <div class="line"><a name="l00236"></a><span class="lineno"> 236</span> CREATE OR REPLACE FUNCTION </div> |
| <div class="line"><a name="l00237"></a><span class="lineno"> 237</span> MADLIB_SCHEMA.plda_cword_count(mystate int4[], doc int4[], topics int4[], doclen int4, num_topics int4, dsize int4)</div> |
| <div class="line"><a name="l00238"></a><span class="lineno"> 238</span> RETURNS int4[]</div> |
| <div class="line"><a name="l00239"></a><span class="lineno"> 239</span> AS <span class="stringliteral">'MODULE_PATHNAME'</span>, <span class="stringliteral">'cword_count'</span> LANGUAGE C;</div> |
| <div class="line"><a name="l00240"></a><span class="lineno"> 240</span> </div> |
| <div class="line"><a name="l00241"></a><span class="lineno"> 241</span> -- Aggregate <span class="keyword">function</span> to compute all word-topic counts given topic assignments <span class="keywordflow">for each</span> document</div> |
| <div class="line"><a name="l00242"></a><span class="lineno"> 242</span> CREATE AGGREGATE MADLIB_SCHEMA.plda_cword_agg(int4[], int4[], int4, int4, int4) (</div> |
| <div class="line"><a name="l00243"></a><span class="lineno"> 243</span>  sfunc = MADLIB_SCHEMA.plda_cword_count,</div> |
| <div class="line"><a name="l00244"></a><span class="lineno"> 244</span>  stype = int4[] </div> |
| <div class="line"><a name="l00245"></a><span class="lineno"> 245</span> );</div> |
| <div class="line"><a name="l00246"></a><span class="lineno"> 246</span> </div> |
| <div class="line"><a name="l00247"></a><span class="lineno"> 247</span> -- The main parallel LDA learning <span class="keyword">function</span></div> |
| <div class="line"><a name="l00248"></a><span class="lineno"> 248</span> CREATE OR REPLACE FUNCTION</div> |
| <div class="line"><a name="l00249"></a><span class="lineno"> 249</span> MADLIB_SCHEMA.plda_train(num_topics int4, num_iter int4, alpha <span class="keywordtype">float</span>, eta <span class="keywordtype">float</span>, </div> |
| <div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  data_table text, dict_table text, model_table text, output_data_table text) </div> |
| <div class="line"><a name="l00251"></a><span class="lineno"> 251</span> RETURNS int4 AS $$</div> |
| <div class="line"><a name="l00252"></a><span class="lineno"> 252</span> </div> |
| <div class="line"><a name="l00253"></a><span class="lineno"> 253</span>  PythonFunctionBodyOnly(`plda<span class="stringliteral">', `plda'</span>)</div> |
| <div class="line"><a name="l00254"></a><span class="lineno"> 254</span>  </div> |
| <div class="line"><a name="l00255"></a><span class="lineno"> 255</span>  <span class="preprocessor"># schema_madlib comes from PythonFunctionBodyOnly</span></div> |
| <div class="line"><a name="l00256"></a><span class="lineno"> 256</span> <span class="preprocessor"></span> <span class="keywordflow">return</span> plda.plda_train( schema_madlib, num_topics, num_iter, alpha, eta, data_table, dict_table, model_table, output_data_table)</div> |
| <div class="line"><a name="l00257"></a><span class="lineno"> 257</span> </div> |
| <div class="line"><a name="l00258"></a><span class="lineno"> 258</span> $$ LANGUAGE plpythonu;</div> |
| <div class="line"><a name="l00259"></a><span class="lineno"> 259</span> </div> |
| <div class="line"><a name="l00260"></a><span class="lineno"> 260</span> CREATE TYPE MADLIB_SCHEMA.plda_word_weight AS ( word text, prob <span class="keywordtype">float</span>, wcount int4 );</div> |
| <div class="line"><a name="l00261"></a><span class="lineno"> 261</span> </div> |
| <div class="line"><a name="l00262"></a><span class="lineno"> 262</span> -- Returns the most important words <span class="keywordflow">for each</span> topic, base on Pr( word | topic ).</div> |
| <div class="line"><a name="l00263"></a><span class="lineno"> 263</span> CREATE OR REPLACE FUNCTION </div> |
| <div class="line"><a name="l00264"></a><span class="lineno"> 264</span> MADLIB_SCHEMA.plda_topic_word_prob(num_topics int4, topic int4, model_table text, dict_table text) </div> |
| <div class="line"><a name="l00265"></a><span class="lineno"> 265</span> RETURNS SETOF MADLIB_SCHEMA.plda_word_weight AS $$</div> |
| <div class="line"><a name="l00266"></a><span class="lineno"> 266</span> </div> |
| <div class="line"><a name="l00267"></a><span class="lineno"> 267</span>  PythonFunctionBodyOnly(`plda<span class="stringliteral">', `plda'</span>)</div> |
| <div class="line"><a name="l00268"></a><span class="lineno"> 268</span>  </div> |
| <div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  <span class="preprocessor"># schema_madlib comes from PythonFunctionBodyOnly</span></div> |
| <div class="line"><a name="l00270"></a><span class="lineno"> 270</span> <span class="preprocessor"></span> <span class="keywordflow">return</span> plda.plda_topic_word_prob( schema_madlib, num_topics, topic, model_table, dict_table) </div> |
| <div class="line"><a name="l00271"></a><span class="lineno"> 271</span> </div> |
| <div class="line"><a name="l00272"></a><span class="lineno"> 272</span> $$ LANGUAGE plpythonu;</div> |
| <div class="line"><a name="l00273"></a><span class="lineno"> 273</span> </div> |
| <div class="line"><a name="l00274"></a><span class="lineno"> 274</span> </div> |
| <div class="line"><a name="l00275"></a><span class="lineno"> 275</span> CREATE TYPE MADLIB_SCHEMA.plda_word_distrn AS ( word text, distrn int4[], prob float8[] );</div> |
| <div class="line"><a name="l00276"></a><span class="lineno"> 276</span> </div> |
| <div class="line"><a name="l00277"></a><span class="lineno"> 277</span> -- This <span class="keyword">function</span> computes the topic assignments to words in a document given previously computed</div> |
| <div class="line"><a name="l00278"></a><span class="lineno"> 278</span> -- statistics from the training corpus. </div> |
| <div class="line"><a name="l00279"></a><span class="lineno"> 279</span> -- This <span class="keyword">function</span> has not been rewritten in PL/Python because it is hard to handle structured types</div> |
| <div class="line"><a name="l00280"></a><span class="lineno"> 280</span> -- like MADLIB_SCHEMA.plda_topics_t in PL/Python (<span class="keyword">this</span> has to be done via conversion to text, which is not nice).</div> |
| <div class="line"><a name="l00281"></a><span class="lineno"> 281</span> CREATE OR REPLACE FUNCTION </div> |
| <div class="line"><a name="l00282"></a><span class="lineno"> 282</span> MADLIB_SCHEMA.plda_label_document(doc int4[], global_count int4[], topic_counts int4[], num_topics int4, dsize int4, </div> |
| <div class="line"><a name="l00283"></a><span class="lineno"> 283</span>  alpha <span class="keywordtype">float</span>, eta <span class="keywordtype">float</span>)</div> |
| <div class="line"><a name="l00284"></a><span class="lineno"> 284</span> RETURNS MADLIB_SCHEMA.plda_topics_t AS $$</div> |
| <div class="line"><a name="l00285"></a><span class="lineno"> 285</span> DECLARE</div> |
| <div class="line"><a name="l00286"></a><span class="lineno"> 286</span>  ret MADLIB_SCHEMA.plda_topics_t;</div> |
| <div class="line"><a name="l00287"></a><span class="lineno"> 287</span> BEGIN</div> |
| <div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  ret := MADLIB_SCHEMA.plda_random_topics(array_upper(doc,1), num_topics);</div> |
| <div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  FOR i in 1..20 LOOP</div> |
| <div class="line"><a name="l00290"></a><span class="lineno"> 290</span>  ret := MADLIB_SCHEMA.plda_sample_new_topics(doc,(ret).topics,(ret).topic_d,global_count,topic_counts,num_topics,dsize,alpha,eta);</div> |
| <div class="line"><a name="l00291"></a><span class="lineno"> 291</span>  END LOOP;</div> |
| <div class="line"><a name="l00292"></a><span class="lineno"> 292</span>  RETURN ret;</div> |
| <div class="line"><a name="l00293"></a><span class="lineno"> 293</span> END;</div> |
| <div class="line"><a name="l00294"></a><span class="lineno"> 294</span> $$ LANGUAGE plpgsql;</div> |
| <div class="line"><a name="l00295"></a><span class="lineno"> 295</span> </div> |
| <div class="line"><a name="l00296"></a><span class="lineno"> 296</span> -- This <span class="keyword">function</span> computes the topic assignments to documents in a test corpus.</div> |
| <div class="line"><a name="l00297"></a><span class="lineno"> 297</span> -- The data_table argument appears unnecessary, as <span class="keywordtype">long</span> as topic_counts is saved in a table from the plda_train() routine<span class="comment"></span></div> |
| <div class="line"><a name="l00298"></a><span class="lineno"> 298</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00299"></a><span class="lineno"> 299</span> <span class="comment"> * @brief This function computes the topic assignments to documents in a test corpus.</span></div> |
| <div class="line"><a name="l00300"></a><span class="lineno"> 300</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00301"></a><span class="lineno"> 301</span> <span class="comment"> * @param test_table Name of table containing the test corpus (must have columns <tt>id INT</tt> and <tt>contents INT[]</tt>).</span></div> |
| <div class="line"><a name="l00302"></a><span class="lineno"> 302</span> <span class="comment"> * @param output_table Name of table to store the results.</span></div> |
| <div class="line"><a name="l00303"></a><span class="lineno"> 303</span> <span class="comment"> * @param model_table Name of table where learned model is stored (in the form of word-topic counts and topic counts).</span></div> |
| <div class="line"><a name="l00304"></a><span class="lineno"> 304</span> <span class="comment"> * @param dict_table Name of the table containing the dictionary (mandatory column is dict text[])</span></div> |
| <div class="line"><a name="l00305"></a><span class="lineno"> 305</span> <span class="comment"> * @param num_topics Number of topics to discover. </span></div> |
| <div class="line"><a name="l00306"></a><span class="lineno"> 306</span> <span class="comment"> * @param alpha Parameter to the topic Dirichlet prior.</span></div> |
| <div class="line"><a name="l00307"></a><span class="lineno"> 307</span> <span class="comment"> * @param eta Parameter to the Dirichlet prior on the per-topic word distributions.</span></div> |
| <div class="line"><a name="l00308"></a><span class="lineno"> 308</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00309"></a><span class="lineno"> 309</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00310"></a><span class="lineno"> 310</span> CREATE OR REPLACE FUNCTION</div> |
| <div class="line"><a name="l00311"></a><span class="lineno"> 311</span> MADLIB_SCHEMA.<a class="code" href="plda_8sql__in.html#adefc4576799e950baf03e7ba66c34eba" title="This function computes the topic assignments to documents in a test corpus. ">plda_label_test_documents</a>(test_table text, output_table text, model_table text, dict_table text, num_topics int4, alpha <span class="keywordtype">float</span>, eta <span class="keywordtype">float</span>)</div> |
| <div class="line"><a name="l00312"></a><span class="lineno"> 312</span> RETURNS VOID AS $$</div> |
| <div class="line"><a name="l00313"></a><span class="lineno"> 313</span> </div> |
| <div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  PythonFunctionBodyOnly(`plda', `plda')</div> |
| <div class="line"><a name="l00315"></a><span class="lineno"> 315</span>  </div> |
| <div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  <span class="preprocessor"># schema_madlib comes from PythonFunctionBodyOnly</span></div> |
| <div class="line"><a name="l00317"></a><span class="lineno"> 317</span> <span class="preprocessor"></span> plda.plda_label_test_documents( schema_madlib, test_table, output_table, model_table, dict_table, num_topics, alpha, eta)</div> |
| <div class="line"><a name="l00318"></a><span class="lineno"> 318</span> </div> |
| <div class="line"><a name="l00319"></a><span class="lineno"> 319</span> $$ LANGUAGE plpythonu;</div> |
| <div class="line"><a name="l00320"></a><span class="lineno"> 320</span> </div> |
| <div class="line"><a name="l00321"></a><span class="lineno"> 321</span> -- Returns the distribution of topics <span class="keywordflow">for</span> a given word</div> |
| <div class="line"><a name="l00322"></a><span class="lineno"> 322</span> CREATE OR REPLACE FUNCTION MADLIB_SCHEMA.plda_word_topic_distrn(arr int4[], ntopics int4, word int4, OUT ret int4[]) AS $$</div> |
| <div class="line"><a name="l00323"></a><span class="lineno"> 323</span>  SELECT $1[(($3-1)*$2 + 1):(($3-1)*$2 + $2)];</div> |
| <div class="line"><a name="l00324"></a><span class="lineno"> 324</span> $$ LANGUAGE sql;<span class="comment"></span></div> |
| <div class="line"><a name="l00325"></a><span class="lineno"> 325</span> <span class="comment">/**</span></div> |
| <div class="line"><a name="l00326"></a><span class="lineno"> 326</span> <span class="comment"> * @brief Main plda function</span></div> |
| <div class="line"><a name="l00327"></a><span class="lineno"> 327</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00328"></a><span class="lineno"> 328</span> <span class="comment"> * @param datatable Name of table containing the corpus (must have columns <tt>id INT</tt> and <tt>contents INT[]</tt>).</span></div> |
| <div class="line"><a name="l00329"></a><span class="lineno"> 329</span> <span class="comment"> * @param dicttable Name of table containing the dictionary (must have column \c dict \c TEXT[]).</span></div> |
| <div class="line"><a name="l00330"></a><span class="lineno"> 330</span> <span class="comment"> * @param modeltable Name of table where learned model will be stored (in the form of word-topic counts and topic counts).</span></div> |
| <div class="line"><a name="l00331"></a><span class="lineno"> 331</span> <span class="comment"> * @param outputdatatable Name of the table the system will store a copy of the datatable plus topic assignments.</span></div> |
| <div class="line"><a name="l00332"></a><span class="lineno"><a class="code" href="plda_8sql__in.html#adefc4576799e950baf03e7ba66c34eba"> 332</a></span> <span class="comment"> * @param numiter Number of iterations to run the Gibbs sampling.</span></div> |
| <div class="line"><a name="l00333"></a><span class="lineno"> 333</span> <span class="comment"> * @param numtopics Number of topics to discover. </span></div> |
| <div class="line"><a name="l00334"></a><span class="lineno"> 334</span> <span class="comment"> * @param alpha Parameter to the topic Dirichlet prior.</span></div> |
| <div class="line"><a name="l00335"></a><span class="lineno"> 335</span> <span class="comment"> * @param eta Parameter to the Dirichlet prior on the per-topic word distributions.</span></div> |
| <div class="line"><a name="l00336"></a><span class="lineno"> 336</span> <span class="comment"> *</span></div> |
| <div class="line"><a name="l00337"></a><span class="lineno"> 337</span> <span class="comment"> */</span></div> |
| <div class="line"><a name="l00338"></a><span class="lineno"> 338</span> CREATE OR REPLACE FUNCTION </div> |
| <div class="line"><a name="l00339"></a><span class="lineno"> 339</span> MADLIB_SCHEMA.plda_run(datatable text, dicttable text, modeltable text, outputdatatable text, </div> |
| <div class="line"><a name="l00340"></a><span class="lineno"> 340</span>  numiter int4, numtopics int4, alpha <span class="keywordtype">float</span>, eta <span class="keywordtype">float</span>)</div> |
| <div class="line"><a name="l00341"></a><span class="lineno"> 341</span> RETURNS VOID AS $$</div> |
| <div class="line"><a name="l00342"></a><span class="lineno"> 342</span> </div> |
| <div class="line"><a name="l00343"></a><span class="lineno"> 343</span>  PythonFunctionBodyOnly(`plda<span class="stringliteral">', `plda'</span>)</div> |
| <div class="line"><a name="l00344"></a><span class="lineno"> 344</span>  </div> |
| <div class="line"><a name="l00345"></a><span class="lineno"> 345</span>  <span class="preprocessor"># schema_madlib comes from PythonFunctionBodyOnly</span></div> |
| <div class="line"><a name="l00346"></a><span class="lineno"> 346</span> <span class="preprocessor"></span> plda.plda_run( schema_madlib, datatable, dicttable, modeltable, outputdatatable, numiter, numtopics, alpha, eta) </div> |
| <div class="line"><a name="l00347"></a><span class="lineno"> 347</span>  </div> |
| <div class="line"><a name="l00348"></a><span class="lineno"> 348</span> $$ LANGUAGE plpythonu;</div> |
| <div class="line"><a name="l00349"></a><span class="lineno"> 349</span> </div> |
| <div class="line"><a name="l00350"></a><span class="lineno"> 350</span> </div> |
| </div><!-- fragment --></div><!-- contents --> |
| </div><!-- doc-content --> |
| <!-- start footer part --> |
| <div id="nav-path" class="navpath"><!-- id is needed for treeview function! --> |
| <ul> |
| <li class="navelem"><a class="el" href="dir_704eb8350b43e1ca74c0f90ed1ba450e.html">methods</a></li><li class="navelem"><a class="el" href="dir_cadc2f84a45765297e3de71a0990944f.html">plda</a></li><li class="navelem"><a class="el" href="dir_e10a5423302f08b62a9f40fa9e226bb0.html">src</a></li><li class="navelem"><a class="el" href="dir_d2147977cd6306f191f6bf61818cc41e.html">pg_gp</a></li><li class="navelem"><a class="el" href="plda_8sql__in.html">plda.sql_in</a></li> |
| <li class="footer">Generated on Tue Sep 10 2013 13:36:13 for MADlib by |
| <a href="http://www.doxygen.org/index.html"> |
| <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.4 </li> |
| </ul> |
| </div> |
| </body> |
| </html> |
