<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article
  PUBLIC "-//NLM//DTD Journal Publishing DTD v2.0 20040830//EN" "http://dtd.nlm.nih.gov/publishing/2.0/journalpublishing.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" article-type="research-article" dtd-version="2.0" xml:lang="EN">

  <front>

    <journal-meta>

      <journal-title>Information Technology Journal</journal-title>

      <issn pub-type="ppub">1812-5638</issn>

      <issn pub-type="epub">1812-5646</issn>

      <publisher>

        <publisher-name>Asian Network for Scientific Information</publisher-name>

      </publisher>

    </journal-meta>


    <article-meta>

      <article-id pub-id-type="doi">10.3923/itj.2009.764.769</article-id>


      <title-group>

        <article-title><![CDATA[Weight-Based Feature Selection for Conditional Maximum Entropy Models]]></article-title>

      </title-group>


      <contrib-group>

        <contrib contrib-type="author" xlink:type="simple">


          <name name-style="western">

            <surname>Li</surname>

            <given-names>Lu</given-names>

          </name>


          <name name-style="western">

            <surname>Wang</surname>

            <given-names>Xuan</given-names>

          </name>


          <name name-style="western">

            <surname>Wang</surname>

            <given-names>XiaoLong</given-names>

          </name>


        </contrib>

      </contrib-group>


      <pub-date pub-type="collection">


        <month>5</month>




        <year>2009</year>

      </pub-date>


      <volume>8</volume>

      <issue>5</issue>


      <abstract><![CDATA[<p>Conditional maximum entropy models provide a unified framework to integrate arbitrary features from different knowledge sources and have been successfully applied to many natural language processing tasks. Feature selection methods are often used to distinguish good features from bad ones to improve model performance. The selection of features in traditional methods is often performed based on different strategies before or along with feature weight estimation, however, weights themselves should be the only factor to measure the importance of features. This study proposes a new selection method based on divide-and-conquer strategies and well-trained feature spaces of small sizes. Features are divided into small subsets, on each of which a sub-model is built and its features are judged according to their weights. The final model is constructed based on merged feature space from all sub-models. Experiments on part of speech tagging show that this method is feasible and efficient.</p>]]></abstract>


    </article-meta>

  </front>


  <ref-list>










      <ref id="6626">

        <label>1</label>

        <citation citation-type="conference" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Berger, A.L. and H. Printz,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1998</year>

          <article-title><![CDATA[A Comparison of criteria for maximum entropy/minimum divergence feature selection.]]></article-title>

          <source>Proceedings of the 3rd Conference on Empirical Methods in Natural Language Processing (EMNLP),</source>

          <volume>1998</volume>

          <fpage>pp: 97</fpage>

          <lpage>106</lpage>

        </citation>

      </ref>












      <ref id="173035">

        <label>2</label>

        <citation citation-type="journal" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Berger, A.L., A. Stephen, Y. Della Pietra, J. Vincent and Y. Della Pietra,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1996</year>

          <article-title><![CDATA[A maximum entropy approach to natural language processing.]]></article-title>

          <source>Comput. Linguistics,</source>

          <volume>22</volume>

          <fpage>39</fpage>

          <lpage>71</lpage>

        </citation>

      </ref>


















      <ref id="173036">

        <label>3</label>

        <citation citation-type="journal" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Chen, Y. and D. Liginlal,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>2008</year>

          <article-title><![CDATA[A maximum entropy approach to feature selection in knowledge-based authentication.]]></article-title>

          <source>Decision Support Syst.,</source>

          <volume>46</volume>

          <fpage>388</fpage>

          <lpage>398</lpage>

        </citation>

      </ref>
























      <ref id="6627">

        <label>4</label>

        <citation citation-type="conference" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Chen, S.F. and R. Rosenfeld,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1999</year>

          <article-title><![CDATA[Efficient sampling and feature selection in whole sentence maximum entropy language models.]]></article-title>

          <source>Proceedings of the Acoustics, Speech and Signal Processing, 1999 IEEE International Conference on ICASSP'99,</source>

          <volume>1999</volume>

          <fpage>pp: 549</fpage>

          <lpage>552</lpage>

        </citation>

      </ref>














      <ref id="16423">

        <label>5</label>

        <citation citation-type="book" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Knuth, D.,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1997</year>

          <article-title><![CDATA[The Art of Computer Programming: Sorting and Searching.]]></article-title>

          <source>The Art of Computer Programming: Sorting and Searching.</source>

          <volume>3rd Edn., Vol. 3,</volume>

          <fpage></fpage>

          <lpage></lpage>

        </citation>

      </ref>






















      <ref id="6628">

        <label>6</label>

        <citation citation-type="conference" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Koeling, R.,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>2000</year>

          <article-title><![CDATA[Chunking with maximum entropy models.]]></article-title>

          <source>Proceedings of the CoNLL-2000 and LLL,</source>

          <volume>2000</volume>

          <fpage>pp: 139</fpage>

          <lpage>141</lpage>

        </citation>

      </ref>












      <ref id="24337">

        <label>7</label>

        <citation citation-type="journal" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Liu, D.C. and J. Nocedal,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1989</year>

          <article-title><![CDATA[On the limited memory BFGS method for large scale optimization.]]></article-title>

          <source>Math Program.,</source>

          <volume>45</volume>

          <fpage>503</fpage>

          <lpage>528</lpage>

        </citation>

      </ref>
























      <ref id="6629">

        <label>8</label>

        <citation citation-type="conference" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Surdeanu, M., R. Johansson, A. Meyers, L. Marquez and J. Nivre,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>2008</year>

          <article-title><![CDATA[The CoNLL-2008 shared task on joint parsing of syntactic and semantic dependencies.]]></article-title>

          <source>Proceedings of the 12th Conference on Computational Natural Language Learning,</source>

          <volume>2008</volume>

          <fpage>pp: 159</fpage>

          <lpage>177</lpage>

        </citation>

      </ref>


















      <ref id="6630">

        <label>9</label>

        <citation citation-type="conference" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Yang, Y.M. and O. Jan Pedersen,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>1997</year>

          <article-title><![CDATA[A comparative study of feature selection in text categorization.]]></article-title>

          <source>Proceedings of the 14th International Conference on Machine Learning,</source>

          <volume>1997</volume>

          <fpage>pp: 412</fpage>

          <lpage>420</lpage>

        </citation>

      </ref>












      <ref id="173249">

        <label>10</label>

        <citation citation-type="journal" xlink:type="simple">

          <person-group person-group-type="author">

            <name name-style="western">

              <surname>Zhou, Y.Q. and L.D. Wu,</surname>

              <given-names></given-names>

            </name>

          </person-group>

          <year>2003</year>

          <article-title><![CDATA[A fast algorithm for feature selection in conditional maximum entropy modeling.]]></article-title>

          <source>Theoretical Issues Natl. Language Process.,</source>

          <volume>10</volume>

          <fpage>153</fpage>

          <lpage>159</lpage>

        </citation>

      </ref>
















  </ref-list>

</article>

