Another usage of provenance data is to verify, replicate or re-apply analysis sessions. Here, we consider the body of work that goes beyond action recovery such as undo/redo. This category of research uses interaction logs to perform real-time or post-hoc quantification to validate the analysis results or to replicate the process when a similar problem arises. For example, in VisTrails an analyst can create, edit, and compare the results of provenance dataflows. The Harvest system tracks interactions with data elements and recommends both notes and relevant views based on previous analyses in a collaborative environment. It is also common to convert the user interactions into executable scripts using a process called program synthesis – generating a script or executable sets of operations. Wrangler, for example, creates data transformation scripts based on passive observations of the user interactions. The scripts can then be re-applied to similar datasets. Knowledge- Pearls allows users to rank and retrieve previous visualization states by formulating a string-based query. The query operates on a provenance graph containing automatically recorded user interactions and visualizations.