Effective multi-label active learning for text classification Export

@inproceedings{citeulike:5123277, abstract = {Labeling text data is quite time-consuming but essential for automatic text classification. Especially, manually creating multiple labels for each document may become impractical when a very large amount of data is needed for training multi-label text classifiers. To minimize the human-labeling efforts, we propose a novel multi-label active learning approach which can reduce the required labeled data without sacrificing the classification accuracy. Traditional active learning algorithms can only handle single-label problems, that is, each data is restricted to have one label. Our approach takes into account the multi-label information, and select the unlabeled data which can lead to the largest reduction of the expected model loss. Specifically, the model loss is approximated by the size of version space, and the reduction rate of the size of version space is optimized with Support Vector Machines (SVM). An effective label prediction method is designed to predict possible labels for each unlabeled data point, and the expected loss for multi-label data is approximated by summing up losses on all labels according to the most confident result of label prediction. Experiments on several real-world data sets (all are publicly available) demonstrate that our approach can obtain promising classification result with much fewer labeled data than state-of-the-art methods.}, address = {New York, NY, USA}, author = {Yang, Bishan and Sun, Jian T. and Wang, Tengjiao and Chen, Zheng}, booktitle = {KDD '09: Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining}, citeulike-article-id = {5123277}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1557019.1557119}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1557019.1557119}, doi = {10.1145/1557019.1557119}, isbn = {978-1-60558-495-9}, keywords = {active-learning, multi-label, text-classification}, location = {Paris, France}, pages = {917--926}, posted-at = {2009-08-11 02:25:16}, priority = {2}, publisher = {ACM}, title = {Effective multi-label active learning for text classification}, url = {http://dx.doi.org/10.1145/1557019.1557119}, year = {2009} }

TY - CONF ID - citeulike:5123277 L3 - citeulike-article-id:5123277 N2 - Labeling text data is quite time-consuming but essential for automatic text classification. Especially, manually creating multiple labels for each document may become impractical when a very large amount of data is needed for training multi-label text classifiers. To minimize the human-labeling efforts, we propose a novel multi-label active learning approach which can reduce the required labeled data without sacrificing the classification accuracy. Traditional active learning algorithms can only handle single-label problems, that is, each data is restricted to have one label. Our approach takes into account the multi-label information, and select the unlabeled data which can lead to the largest reduction of the expected model loss. Specifically, the model loss is approximated by the size of version space, and the reduction rate of the size of version space is optimized with Support Vector Machines (SVM). An effective label prediction method is designed to predict possible labels for each unlabeled data point, and the expected loss for multi-label data is approximated by summing up losses on all labels according to the most confident result of label prediction. Experiments on several real-world data sets (all are publicly available) demonstrate that our approach can obtain promising classification result with much fewer labeled data than state-of-the-art methods. CY - Paris, France SN - 978-1-60558-495-9 AD - New York, NY, USA EP - 926 TI - Effective multi-label active learning for text classification PB - ACM T2 - KDD '09: Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining SP - 917 KW - active-learning KW - multi-label KW - text-classification AU - Yang, Bishan AU - Sun, Jian AU - Wang, Tengjiao AU - Chen, Zheng PY - 2009/// UR - http://dx.doi.org/10.1145/1557019.1557119 ER -