International Center for Ethics in the Sciences and Humanities (IZEW)
DQ-Navigator
The DQ Navigator serves as a working tool for handling and improving data quality (DQ) for artificial intelligence applications. By independently answering and examining selected, scientifically based questions, users can question and discuss the quality of individual data sets and identify potential for improvement independently.
The content of the questionnaire is based on a meta-analysis of various acknowledged scientific publications that address the requirements for data (sets) from different perspectives. In the course of the meta-analysis, DQ dimensions (e.g. diversity, credibility, representativeness, security and transparency) were also identified, which enable different perspectives on data quality. Each question can be assigned to specific dimensions relevant to the data set. The DQ Navigator offers the opportunity to draw conclusions about the quality of a specific data set or to examine DQ in general by working through the questionnaire.
The centrepiece of the DQ Navigator is a catalogue of 126 questions for independent data quality (DQ) assessment. The questions are intended as suggestions for examining specific quality aspects of your own data set. The ‘Dimensions’ tab explains the concept of dimensions, which is important for the questionnaire, and presents the DQ-relevant dimensions identified by the research project. Each question in the catalogue can be assigned to one or more dimensions relevant to the data set and can be filtered according to those dimensions. The dimensions of the questionnaire are also divided into main and secondary dimensions. If the dimensions or identical terms have already been taken up in legislation, a corresponding note is provided.
To get an overview of how many questions have been addressed in total and how many relate to specific dimensions, statistics are available in the ‘Progress’ tab.
Given the complexity of the subject matter, independent use of the DQ Navigator cannot provide conclusive assessments of individual data sets and therefore cannot produce concrete work results. The questionnaire is not exhaustive and only includes the sources used by the research consortium in the meta-analysis. It is possible to expand the catalogue to include additional questions relevant to data quality. There is no external evaluation of your own assessment. The use of the DQ Navigator does not allow any reliable conclusions to be drawn about compliance with legal requirements.
