**1. Introduction**

Classification is the most popular task in Data Mining. It consists of attribute to the appropriate class to an instance. There are several fields of classification that depend on the number of classes and the number of possible values of a class in a dataset. If a dataset contains a single class, which can have two values, then we speak about the classification in binary datasets. However, if the single class has more than two values, then we speak about the classification in multi-class datasets. In a case where a dataset contains several classes at a time, we speak about the classification in multi-label datasets.

Multi-label datasets appear in several applications such as text categorization [1], image annotation [2, 3], web advertising [4], and music categorization [5]. In these applications there are usually tens or hundreds of thousands of labels, while the number is still increasing. It is important to extract knowledge from these datasets to take decision. Consequently, the problem of classification in this kind of datasets is being an important problem in machine learning, and it has attracted the attention of many researchers.

For supervised learning algorithm from multi-label datasets, there are two major tasks: multi-label classification MLC and label ranking (LR) [6]. The first task is concerned with learning a model that outputs a bipartition of labels into relevant and irrelevant labels. The second task is concerned with learning model that outputs an ordering of the class labels according to their relevance.

For both tasks, the different approaches and techniques are proposed to deal with the classification in multi-label datasets divided into the two categories: problem transformation methods and algorithm adaptation methods [7]. In the first category, multi-label classification problem is transformed into one or more single classification problems. However, in the second one, the existing approaches are adapted to the studied problem.

This chapter is organized as follows: Section 2 presents the different notations used in the rest of chapter. In Section 3, we present at first the description measures of a multi-label dataset, and in the Section 4, we present the evaluation metrics used to evaluate the performances of the test dataset. In Section 5, we detail the different approaches and techniques used to deal with the problem of classification in multi-label datasets. Finally, in Section 6, we make our concluding remarks.
