The goal of visualization is to create visual representations to support the user’s reasoning and decisionmaking with data. Consequently, one of the primary reasons for analyzing provenance data is to understand the user and their sensemaking process. The ultimate goal of this category of research is to create theoretical and computational models that can describe the human analytical reasoning process. Some of the earlier research in the area works to uncover analysis patterns from interaction log data. For example, Dou et al. demonstrated that it is possible to recover analysts’ findings and strategies from log data. More recent work uses computational methods to uncover analysis patterns and workflows. A promising set of work has also started to learn individual user characteristics, such as expertise, personality traits, and cognitive abilities from provenance data. Also in this category is work on modeling attention and exploration biases during analysis.