This chapter examines the foundational principles of learning ranking functions. It begins by introducing specialized evaluation metrics that are essential for assessing the quality of ranked outputs. These metrics are designed to account for the positions of relevant instances within a ranked list, offering a nuanced measure of a model’s ranking effectiveness. The chapter distinguishes between pointwise and pairwise approaches to the ranking problem. Pointwise methods treat ranking as a regression or classification task for individual items, while pairwise approaches focus on learning the relative order between pairs of instances, effectively transforming the ranking task into a binary classification problem on pairs. This pairwise methodology introduces dependencies between examples, as pairs are inherently linked, necessitating new learning frameworks capable of handling such interdependent data. The chapter also addresses ongoing challenges in the field, including scaling to large datasets. Through this comprehensive exploration, the chapter provides a solid foundation for understanding both the theoretical and practical aspects of Learning-to-Rank.