EnTrust (Teilprojekt 3: Erklärbarkeit)

The goal of the EnTrust research focus is to explore methodological approaches and formal tools that support the engineering development of correct - or at least trustworthy - data-intensive software. This goal is pursued in various subprojects.

A key aspect that can contribute to the trustworthiness of data-intensive systems is their explainability, i.e., their ability to explain decisions made, but also, more generally, to prove or demonstrate the correct execution of such systems. Although in data-intensive systems "correct execution" is often not clearly defined, there are aspects of "correctness" that can be demonstrated for.

(i) Data,

(ii) algorithms that process data, and

(iii) decisions based on such algorithms.

The goal of Subproject 3 (Explanability)is to develop methods to provide explanations for the correctness and reper- sentativity of data, as well as correctness and fairness of algorithms that process such data, and to explain decisions based on data-intensive software.