Spam Filtering

Kleanthi Georgala, Aris Kosmopoulos, George Paliouras
2014 Proceedings of the 4th International Conference on Web Intelligence, Mining and Semantics (WIMS14) - WIMS '14  
 Unsupervised learning, based on local structure  Create groups of highly correlated data-points  No re-clustering of data  Our contribution : Active Learning combined with Incremental Clustering  Use only 2% of the overall message labels  Consider natural grouping of data : select representative instances for training
doi:10.1145/2611040.2611059 dblp:conf/wims/GeorgalaKP14 fatcat:qcet2r7jazgdtddp7qvrwa5acy