Semi-supervised clustering with user feedback Export

@misc{citeulike:267343, abstract = {We present a new approach to clustering based on the observation that \it is easier to criticize than to construct.\" Our approach of semi-supervised clustering allows a user to iteratively provide feedback to a clustering algorithm. The feedback is incorporated in the form of constraints which the clustering algorithm attempts to satisfy on future iterations. These constraints allow the user to guide the clusterer towards clusterings of the data that the user nds more useful. We...}, author = {Cohn, D. and Caruana, R. and Mccallum, A.}, citeulike-article-id = {267343}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.33.906}, comment = {Provide initial clustering User can now critisize by * this document does not belong to this cluster * move document to a different cluster * these two documents do not belong together They use EM. They have an application where they learnt how important words are for forming the cluster structure.}, keywords = {active, clustering, contraint}, posted-at = {2005-07-28 17:04:17}, priority = {0}, title = {Semi-supervised clustering with user feedback}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.33.906}, year = {2003} }

TY - GEN ID - citeulike:267343 L3 - citeulike-article-id:267343 N2 - We present a new approach to clustering based on the observation that \it is easier to criticize than to construct." Our approach of semi-supervised clustering allows a user to iteratively provide feedback to a clustering algorithm. The feedback is incorporated in the form of constraints which the clustering algorithm attempts to satisfy on future iterations. These constraints allow the user to guide the clusterer towards clusterings of the data that the user nds more useful. We... TI - Semi-supervised clustering with user feedback KW - active KW - clustering KW - contraint N1 - Provide initial clustering User can now critisize by * this document does not belong to this cluster * move document to a different cluster * these two documents do not belong together They use EM. They have an application where they learnt how important words are for forming the cluster structure. AU - Cohn, D AU - Caruana, R AU - Mccallum, A PY - 2003/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.33.906 ER -