Document representations for classification of short web-page descriptions
Motivated by applying Text Categorization to classification of Web search results, this paper describes an extensive experimental study of the impact of bag-of- words document representations on the performance of five major classifiers - Naïve Bayes, SVM, Voted Perceptron, kNN and C4.5. The texts,...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Belgrade
2008-01-01
|
| Series: | Yugoslav Journal of Operations Research |
| Subjects: | |
| Online Access: | http://www.doiserbia.nb.rs/img/doi/0354-0243/2008/0354-02430801123R.pdf |
| _version_ | 1830282957294665728 |
|---|---|
| author | Radovanović Miloš Ivanović Mirjana |
| author_facet | Radovanović Miloš Ivanović Mirjana |
| author_sort | Radovanović Miloš |
| collection | DOAJ |
| description | Motivated by applying Text Categorization to classification of Web search results, this paper describes an extensive experimental study of the impact of bag-of- words document representations on the performance of five major classifiers - Naïve Bayes, SVM, Voted Perceptron, kNN and C4.5. The texts, representing short Web-page descriptions sorted into a large hierarchy of topics, are taken from the dmoz Open Directory Web-page ontology, and classifiers are trained to automatically determine the topics which may be relevant to a previously unseen Web-page. Different transformations of input data: stemming, normalization, logtf and idf, together with dimensionality reduction, are found to have a statistically significant improving or degrading effect on classification performance measured by classical metrics - accuracy, precision, recall, F1 and F2. The emphasis of the study is not on determining the best document representation which corresponds to each classifier, but rather on describing the effects of every individual transformation on classification, together with their mutual relationships. . |
| first_indexed | 2024-12-19T02:48:20Z |
| format | Article |
| id | doaj.art-51a641bc24474d278a0a10de1bdeeb64 |
| institution | Directory Open Access Journal |
| issn | 0354-0243 1820-743X |
| language | English |
| last_indexed | 2024-12-19T02:48:20Z |
| publishDate | 2008-01-01 |
| publisher | University of Belgrade |
| record_format | Article |
| series | Yugoslav Journal of Operations Research |
| spelling | doaj.art-51a641bc24474d278a0a10de1bdeeb642022-12-21T20:38:46ZengUniversity of BelgradeYugoslav Journal of Operations Research0354-02431820-743X2008-01-0118112313810.2298/YJOR0801123RDocument representations for classification of short web-page descriptionsRadovanović MilošIvanović MirjanaMotivated by applying Text Categorization to classification of Web search results, this paper describes an extensive experimental study of the impact of bag-of- words document representations on the performance of five major classifiers - Naïve Bayes, SVM, Voted Perceptron, kNN and C4.5. The texts, representing short Web-page descriptions sorted into a large hierarchy of topics, are taken from the dmoz Open Directory Web-page ontology, and classifiers are trained to automatically determine the topics which may be relevant to a previously unseen Web-page. Different transformations of input data: stemming, normalization, logtf and idf, together with dimensionality reduction, are found to have a statistically significant improving or degrading effect on classification performance measured by classical metrics - accuracy, precision, recall, F1 and F2. The emphasis of the study is not on determining the best document representation which corresponds to each classifier, but rather on describing the effects of every individual transformation on classification, together with their mutual relationships. .http://www.doiserbia.nb.rs/img/doi/0354-0243/2008/0354-02430801123R.pdftext categorizationdocument representationmachine learning |
| spellingShingle | Radovanović Miloš Ivanović Mirjana Document representations for classification of short web-page descriptions Yugoslav Journal of Operations Research text categorization document representation machine learning |
| title | Document representations for classification of short web-page descriptions |
| title_full | Document representations for classification of short web-page descriptions |
| title_fullStr | Document representations for classification of short web-page descriptions |
| title_full_unstemmed | Document representations for classification of short web-page descriptions |
| title_short | Document representations for classification of short web-page descriptions |
| title_sort | document representations for classification of short web page descriptions |
| topic | text categorization document representation machine learning |
| url | http://www.doiserbia.nb.rs/img/doi/0354-0243/2008/0354-02430801123R.pdf |
| work_keys_str_mv | AT radovanovicmilos documentrepresentationsforclassificationofshortwebpagedescriptions AT ivanovicmirjana documentrepresentationsforclassificationofshortwebpagedescriptions |