CP-nets are a graphical representation for qualitative conditional ceteris paribus preference statements, allowing for compact and intuitive specification of preference relations. They capture conditional dependencies and independences among preferences, enabling efficient inference tasks such as determining outcome dominance, ordering outcomes, and constructing the best outcome given partial evidence. The paper introduces CP-nets, which are directed graphs where each node represents a variable and is annotated with a conditional preference table (CPT) that specifies preferences over values of the variable given its parent values. The semantics of CP-nets are defined in terms of preference rankings consistent with the CPTs, and the paper discusses the satisfiability of CP-nets, the implications of ceteris paribus semantics, and the use of CP-nets in outcome optimization. The paper also presents an algorithm for outcome optimization in CP-nets and discusses the computational complexity of dominance and ordering queries. The paper concludes with a discussion of related work and future research directions.CP-nets are a graphical representation for qualitative conditional ceteris paribus preference statements, allowing for compact and intuitive specification of preference relations. They capture conditional dependencies and independences among preferences, enabling efficient inference tasks such as determining outcome dominance, ordering outcomes, and constructing the best outcome given partial evidence. The paper introduces CP-nets, which are directed graphs where each node represents a variable and is annotated with a conditional preference table (CPT) that specifies preferences over values of the variable given its parent values. The semantics of CP-nets are defined in terms of preference rankings consistent with the CPTs, and the paper discusses the satisfiability of CP-nets, the implications of ceteris paribus semantics, and the use of CP-nets in outcome optimization. The paper also presents an algorithm for outcome optimization in CP-nets and discusses the computational complexity of dominance and ordering queries. The paper concludes with a discussion of related work and future research directions.