Auditing complex Artificial Intelligence (AI) systems is gaining importance in light of new regulations and is particularly challenging in terms of system complexity, knowledge integration, and differing transparency needs. Current AI auditing tools however, lack semantic context, resulting in difficulties for auditors in effectively collecting and integrating, but also for analysing and querying audit data. In this position paper, we explore how Knowledge Graphs (KGs) can address these challenges by offering a structured and integrative approach to collecting and transforming audit traces. This work discusses the current limitations in both AI auditing processes and tools. Furthermore, we examine how KGs can play a transformative role in overcoming these obstacles to achieve improved auditability and transparency of AI systems.

Industry 4.0 envisions adaptive production systems, i.e., Cyber-Physical Production Systems (CPPSs), to manufacture products from a product line. Product-Process-Resource modeling represents the essential aspects of a CPPS. However, due to discipline-specific models, e.g., mechanical, electrical, and automation models, it is often unclear how to integrate the proprietary data into an integrated model due to missing common understanding. This paper investigates (i) how to integrate local engineering views with Common Concepts (CCs) and using them as a defined taxonomy for modeling a network of engineering concepts; (ii) how to build an engineering network graph for visualisation and analysis considering discipline-specific needs. We motivate a method to support CPPS engineering organisations to integrate their heterogeneous data using CCs. This builds the basis for defining multi-domain engineering graphs for visualisation and analysis aspects. In this paper, we present a research agenda discussing open issues and expected results.

In many countries worldwide, including Austria, the environmental impact of production facilities is strongly regulated leading to authorities issuing a large number of legal permits on this topic. The access of interested parties to these permits is typically supported by search systems that present a structured view of the permits along their key elements, such as issuing authority or their legal basis. In this paper, we present a real-life use case from Austria's Environment Agency, where the extraction of such key elements represents a non-trivial task for laypersons with limited legal knowledge: the heterogeneity of data, complex language, and implicit information hinder the manual data extraction process and can lead to poor quality in data management. Based on an analysis of the use case's main requirements, we propose an architecture for a system to support the extraction of key elements from legal permits by laypersons. The system combines methods and techniques based on Knowledge Graphs / Semantic Web and Machine Learning technologies and aims to be auditable in terms of its operation.

Technical Debt (TD) has proven to be a suitable communication concept for software-intensive contexts to raise awareness regarding longterm negative effects of deviations from standards and guidelines. TD has also been introduced to systems engineering domain, to communicate design shortcomings in long-running, software-assisted systems. We analysed potential TD in the engineering data exchange for production system engineering. Similar to requirements engineering in software-intensive systems, data exchange in the design phase plays an integral part in Software Engineering (SE) for Production Systems Engineering: Specifications, and physical logic have to be derived from heterogeneous plant models or parameter tables designed by different stakeholders. However, traditional procedures and inadequate tool support lead to inefficient data extraction and integration. We identified debt arising from knowledge representation, data model and the exchange process. The refinement validation of identified TD was achieved through semi-structured interviews with representatives in two analysed companies. In an online survey with ten participants from an industrial consortium we evaluated whether the identified TD concepts also applied to other companies, which is true for the majority of TD. Furthermore, we discuss promising TD management strategies to repay and manage negative effects and the accumulation of additional debt, such as improved communication, test-driven model engineering and visualisation of engineering models.

Multi-disciplinary data exchange still poses many challenges: Heterogeneous data sources, diverging data views, and lack of communication lead to defects, late resolving of errors, and mismatches over the project lifecycle. Semantic approaches such as ontologies are a viable solution to derive common concepts between disciplines to limit negative effects in their collaboration. However, the application of these semantic approaches is still quite limited due to its inherent complexity. The purpose of this paper, is to discuss the concept of local glossaries as a step towards a Common Concept Glossary (CCG) method: A tool-supported method to enable and simplify the creation of common concepts derived from local glossaries that are built by discipline-specific workgroups. The local concepts can aid by making changes visible and enabling traceability and maintainability of common models between different disciplines.

Similar to advanced software engineering, multi-disciplinary systems engineering, such as industrial production systems engineering (PSE), has to integrate partial results from workgroups that design in parallel. Due to the heterogeneity of data sources and the divergence of local data models the exchanged engineering artefacts between PSE workgroups are complex, making the automation of the data exchange process in PSE difficult and prone to technical debt (TD). In this paper, we report on a case study at a large PSE company to analyze TD effects, items, and causes in PSE, focusing on the engineering data exchange process. We identified key use cases and TD types, i.e., TD in data models and TD in data instances, which have adverse effects on project effort, cost and duration as well as data quality.

The realization of a cyber-physical production system (CPPS) requires suitable methods and tools for the exchange and integration of engineering data between collaborating disciplines. Unfortunately, the description languages used in a CPPS Engineering (CPPSE) organization to describe discipline-specific views are not necessarily well suited for high-quality data exchange between workgroups. In this paper, we identify technical debt and risks regarding CPPSE description languages for data exchange using the VDI 3695 guideline as best practice. We report on effects and likely causes of the quality risks identified in a case study at a large CPPSE company. Based on data from workshops and semi-structured interviews with 28 domain experts from 12 workgroups, we propose a preliminary model relating causes and effects as foundation for analyzing and managing risks in the CPPSE data exchange process.

Data exchange and integration in a heterogeneous engineering network are challenges that an engineering organi-zation has to address. Most practitioners have realized that this challenge requires detailed consideration and involvement of the related engineering disciplines to specify and implement an appropriate data logistics. In this paper, we introduce a meta-model and method for the development of a data-mapping infrastructure to support the work of the engineers responsible for data logistics. The meta-model is the foundation for data logistics with automated data transformation, integration, and exchange by automating the mapping of engineering data as envisioned in VDI Guideline 3695. We report on a first validation of the method with a proof of concept using AutomationML.

In the multi-disciplinary production systems engineering (PSE) process, software engineers depend on requirements and design rationales coming from product and production process planning, summarized as product/ion knowledge.Unfortunately, the engineering artifacts coming from product/ion planning often represent important product/ion knowledge incompletely and not well integrated, leading to risks regarding software engineering quality.In this paper, we report on a case study at a large industrial PSE organization, investigating Technical Debt (TD) effects, items, and causes in PSE process documentation and configuration management according to the VDI guideline 3695 Part 2. We focus on requirements for and issues in the representation of product/ion knowledge in the engineering data provided to software engineers.Based on data elicited from PSE domain experts, we model TD concepts based on the Quality Function Deployment method as foundation for TD analysis and risk management.The initial validation with domain experts revealed how software engineers could benefit from improved product/ion knowledge modeling as foundation for better understanding the rationale of engineering design decisions.