Conditional preference networks (CP-nets) exploit the power of ceteris paribus rules to represent preferences over combinatorial decision domains compactly. CP-nets have much appeal. However, their study has not yet advanced sufficiently for their widespread use in real-world applications. Known algorithms for deciding dominance---whether one outcome is better than another with respect to a CP-net---require exponential time. Data for CP-nets are difficult to obtain: human subjects data over combinatorial domains are not readily available, and earlier work on random generation is also problematic. Also, much of the research on CP-nets makes strong, often unrealistic assumptions, such as that decision variables must be binary or that only strict preferences are permitted. In this thesis, I address such limitations to make CP-nets more useful. I show how: to generate CP-nets uniformly randomly; to limit search depth in dominance testing given expectations about sets of CP-nets; and to use local search for learning restricted classes of CP-nets from choice data.

La croissance exponentielle des données personnelles, et leur mise à disposition sur la toile, a motivé l’émergence d’algorithmes d’apprentissage de préférences à des fins de recommandation, ou d’aide à la décision. Les réseaux de préférences conditionnelles (CP-nets) fournissent une structure compacte et intuitive pour la représentation de telles préférences. Cependant, leur nature combinatoire rend leur apprentissage difficile : comment apprendre efficacement un CP-net au sein d’un milieu bruité, tout en supportant le passage à l’échelle ?Notre réponse prend la forme de deux algorithmes d’apprentissage dont l’efficacité est soutenue par de multiples expériences effectuées sur des données réelles et synthétiques.Le premier algorithme se base sur des requêtes posées à des utilisateurs, tout en prenant en compte leurs divergences d’opinions. Le deuxième algorithme, composé d’une version hors ligne et en ligne, effectue une analyse statistique des préférences reçues et potentiellement bruitées. La borne de McDiarmid est en outre utilisée afin de garantir un apprentissage en ligne efficace. / The rapid growth of personal web data has motivated the emergence of learning algorithms well suited to capture users’ preferences. Among preference representation formalisms, conditional preference networks (CP-nets) have proven to be effective due to their compact and explainable structure. However, their learning is difficult due to their combinatorial nature.In this thesis, we tackle the problem of learning CP-nets from corrupted large datasets. Three new algorithms are introduced and studied on both synthetic and real datasets.The first algorithm is based on query learning and considers the contradictions between multiple users’ preferences by searching in a principled way the variables that affect the preferences. The second algorithm relies on information-theoretic measures defined over the induced preference rules, which allow us to deal with corrupted data. An online version of this algorithm is also provided, by exploiting the McDiarmid's bound to define an asymptotically optimal decision criterion for selecting the best conditioned variable and hence allowing to deal with possibly infinite data streams.