The Research On Cllective Multi-Label Classification

Text classification is based on the text will be open to the text into one or more of the pre-defined the method of classification. the text could be better to solve a lot of documents and information collation of the problem can be applied to many things.Text classification include two kinds of single label and multil-label. Classified technology is a label to the current text to a predefined class( the class marked ); label on the contrary, many categories of technologies may be given the current text to a number of predefined clas(s much class marked ..Common approaches to multi-label classification learn independent classifiers for each category, and employ ranking or thresholding schemes for classification. Because they do not exploit dependencies between labels, such techniques are only well-suited to problems in which categories are independent. Multi-Label classifier (CML) also, jointly,learns parameters for each pair of labels. The Collective Multi-Label with Features classifier (CMLF) learns parameters for feature label-label triples--capturing the impact that an individual feature has on the co-occurrence probability of a pair of labels. The Collective Multi-Label with Features classifier (CMLF) learns parameters for feature label-label triples—capturing the impact that an individual feature has on the co-occurrence probability of a pair of labels. This paper explores multilabel conditional random field (CRF) classification models that directly parameterize label co-occurrences in multi-label classification. on the basis of Multi-Label with Features classifier use to support the classification(SVM)classification of text can be improved obviously higher efficiency.