SCGaz - A Synthetic Formal Context Generator with Density Control for Test and Evaluation of FCA Algorithms

An efficient way to evaluate FCA algorithms is through a comparative analysis of their performance in typical contexts. Comparisons are normally conducted using randomly generated contexts that may contain duplicated attributes and objects and other types of redundancies. Failing to acknowledge the presence of these redundancies in formal contexts could lead to erroneous comparison analysis. This paper proposes a tool named SCGaz (Synthetic Context Generator) that randomly fills synthetic formal contexts ensuring the absence of some type of redundancies. At the same time, the tool is able to keep track of the contexts density, allowing users to select any density in the bounds of the minimum and maximum permitted for a type of context. Thus, this approach allows more controllable and reliable simulation environment. In this work, an analysis of the time spent to generate different types of formal contexts, including large ones, is presented. As a case study, a performance comparison between Object Intersection algorithm and its dual version, Attribute Intersections, with contexts generated by SCGaz is discussed. Contexts produced by SCGaz in conjunction with real world dataset allow a more in-depth comparative analysis of FCA algorithms performance.