%0 Journal Article %1 baykan2011comprehensive %A Baykan, Eda %A Henzinger, Monika %A Marian, Ludmila %A Weber, Ingmar %D 2011 %I Association for Computing Machinery (ACM) %J Transactions on the Web %K classification link url web %N 3 %P 1--29 %R 10.1145/1993053.1993057 %T A Comprehensive Study of Features and Algorithms for URL-Based Topic Classification %U https://doi.org/10.1145%2F1993053.1993057 %V 5 %X Given only the URL of a Web page, can we identify its topic? We study this problem in detail by exploring a large number of different feature sets and algorithms on several datasets. We also show that the inherent overlap between topics and the sparsity of the information in URLs makes this a very challenging problem. Web page classification without a page’s content is desirable when the content is not available at all, when a classification is needed before obtaining the content, or when classification speed is of utmost importance. For our experiments we used five different corpora comprising a total of about 3 million (URL, classification) pairs. We evaluated several techniques for feature generation and classification algorithms. The individual binary classifiers were then combined via boosting into metabinary classifiers. We achieve typical F-measure values between 80 and 85, and a typical precision of around 86. The precision can be pushed further over 90 while maintaining a typical level of recall between 30 and 40.

@article{baykan2011comprehensive, abstract = {Given only the URL of a Web page, can we identify its topic? We study this problem in detail by exploring a large number of different feature sets and algorithms on several datasets. We also show that the inherent overlap between topics and the sparsity of the information in URLs makes this a very challenging problem. Web page classification without a page’s content is desirable when the content is not available at all, when a classification is needed before obtaining the content, or when classification speed is of utmost importance. For our experiments we used five different corpora comprising a total of about 3 million (URL, classification) pairs. We evaluated several techniques for feature generation and classification algorithms. The individual binary classifiers were then combined via boosting into metabinary classifiers. We achieve typical F-measure values between 80 and 85, and a typical precision of around 86. The precision can be pushed further over 90 while maintaining a typical level of recall between 30 and 40.}, added-at = {2023-11-08T15:41:06.000+0100}, author = {Baykan, Eda and Henzinger, Monika and Marian, Ludmila and Weber, Ingmar}, biburl = {https://www.bibsonomy.org/bibtex/2383cc06c01b67cadc222039436907886/jaeschke}, description = {A Comprehensive Study of Features and Algorithms for URL-Based Topic Classification | ACM Transactions on the Web}, doi = {10.1145/1993053.1993057}, interhash = {47b728419cbb49db61f8bd4c2871f2ad}, intrahash = {383cc06c01b67cadc222039436907886}, journal = {Transactions on the Web}, keywords = {classification link url web}, month = jul, number = 3, pages = {1--29}, publisher = {Association for Computing Machinery ({ACM})}, timestamp = {2023-11-08T15:41:06.000+0100}, title = {A Comprehensive Study of Features and Algorithms for URL-Based Topic Classification}, url = {https://doi.org/10.1145%2F1993053.1993057}, volume = 5, year = 2011 }