Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
An Experimental Study on Large-Scale Web Categorization

Tie-Yan Liu, Yiming Yang, Hao Wan, Qian Zhou, Bin Gao, Hua-Jun Zeng, Zheng Chen, and Wei-Ying Ma


Taxonomies of the Web typically have hundreds of thousands of categories and skewed category distribution over documents. It is not clear whether existing text classification technologies can perform well on and scale up to such large-scale applications. To understand this, we conducted the evaluation of several representative methods (Support Vector Machines, k-Nearest Neighbor and Naive Bayes) with Yahoo! taxonomies. In particular, we evaluated the effectiveness/efficiency tradeoff in classifiers with hierarchical setting compared to conventional (flat) setting, and tested popular threshold tuning strategies for their scalability and accuracy in large-scale classification problems.


Publication typeArticle
Published inSpecial interest tracks and posters of the 14th international conference on World Wide Web
PublisherAssociation for Computing Machinery, Inc.
> Publications > An Experimental Study on Large-Scale Web Categorization