What is Two-Step Strategy of PU Learning

In traditional text categorization, a classifier is built using labeled training documents from a set of predefined classes. This chapter studies a different problem: partially supervised text categorization. Given a set P of positive documents of a particular class and a set U of unlabeled documents (which contains both hidden positive and hidden negative documents), we build a classifier using P and U to classify the data in U as well as future test data. The key feature of this problem is that there is no labeled negative document, which makes traditional text classification techniques inapplicable. In this chapter, we introduce the main techniques S-EM, PEBL, Roc-SVM and A-EM, to solve the partially supervised problem. In many application domains, partially supervised text categorization is preferred since it saves on the labor-intensive effort of manual labeling of negative documents.