Multi-label learning deals with data associated with multiple labels simultaneously.It has been extensively studied in diverse areas such as information retrieval,bioinformatics,image annotation,etc.Explosive growth of multi-label related data has brought challenges of how to efficiently learn these labeled data and automatically label the unlabeled data.In this paper,we propose an online learning algorithm which processes the data arrived in streaming fashion.It is space-saving and scalable to large-scale problems.Specifically,to tackle the class imbalance problem we exploit label prior to construct cost sensitive function for sub-classification problem.Experimental studies corroborate the performance of our approaches on data sets drawn from diverse domains.