Local Causal Discovery with Background Knowledge

Causality plays a pivotal role in various fields of study. Based on the framework of causal graphical models, previous works have proposed identifying whether a variable is a cause or non-cause of another variable in every Markov equivalent graph by learning only the local structure. However, the presence of prior knowledge, often represented as a partially known causal graph, is common in many causal modeling applications. Leveraging this prior knowledge enables further identification of causal relations. In this paper, we first propose a method for learning the local structure by incorporating several types of causal background knowledge, including direct causal, non-ancestral, and ancestral information. Then we introduce sufficient and necessary conditions for identifying causal relations based solely on the local structure in the presence of prior knowledge. The effectiveness and efficiency of our method are demonstrated through experiments on local structure learning, causal relation identification, and its application to fair machine learning.