Classification Models
Classification Models

The most common technique for evaluating provenance data is the use of classification and statistical modeling techniques to differentiate sequences of user actions. The overarching goal of such techniques is to map a user action to one or more categories. A number of surveys in the literature have demonstrated the application of these techniques to a variety of data types relevant to provenance analysis, including text and images. Indeed, many of the common types of insights that users wish to identify in data necessitate a classification phase, including comparison, correlation, distribution, and trend insights. In this section, we identify publications that classify provenance data into groups of similar user actions via a variety of methods, ranging from straightforward clustering through complex machine learning processes.

Examples Figures from the Literature

Toker et al. 2017
Steichen et al. 2014
Wegba et al. 2018


Example Papers