ZDIN-Zahlen im Detail

Major online platforms today are multi-stakeholder in nature and they cater to the interests of various stakeholders. Examples include e-commerce platforms with sellers of goods and services and buyers who pay for them, hotel booking platforms with hosts and guests, media-streaming platforms artists or content creators and viewers, and many more. The focus here is on the information access services (like search and recommendation) deployed on these platforms. While traditionally these services were designed in consumer-centric ways, they are found to be unfair to providers. Since the providers depend on these platforms for their livelihood, fairness for providers is an important and necessary design element in many scenarios. This chapter summarizes the domain and discusses various prior works on provider fairness in such multi-stakeholder platforms after a brief review of major works on algorithmic fairness in machine learning. Many recent works show that provider fairness can cause loss of utility and unfairness for consumers. Following this, a number of works have proposed fairness notions for both providers and consumers in different settings, and studied multi-stakeholder fairness to balance various fairness and utility goals or constraints. This chapter reviews some major works on multi-stakeholder fairness in the following three aspects: the problem setting, fairness notions, and proposed approaches. It also discusses how most of the works on multi-stakeholder fairness have considered settings with only two stakeholders, gives some examples on other platforms with more than two stakeholders, and how they are fundamentally different in terms of the utility structure for the stakeholders.

Eine zentrale Herausforderung bei der Umsetzung von Model-based Systems Engineering ist die Sicherstellung der Modellkonsistenz. Um dieser Herausforderung zu begegnen, schlägt dieser Beitrag die Anwendung von Konsistenzregeln vor. Hierzu werden einerseits notwendige Grundlagen, wie die Bewertung der Modellkonsistenz und andererseits die konkrete Formulierung von Konsistenzregeln betrachtet. Um eine strukturierte Formulierung zu ermöglichen, wurde eine Satzschablone entwickelt, welche insbesondere die Bewertung von Architekturmustern beinhaltet. Außerdem wurden entsprechend der Satzschablone konkrete Konsistenzregeln an einem heterogenen Modell beispielhaft angewendet.

Abstract:
Irregularly sampled time series data with missing values is a observed in many fields like healthcare, astronomy, and climate science. Interpolation of these types of time series is crucial for tasks such as root cause analysis and medical diagnosis, as well as for smoothing out irregular or noisy data. To address this challenge, we present a novel encoder-decoder architecture called "Tripletformer" for probabilistic interpolation of irregularly sampled time series with missing values. This attention-based model operates on sets of observations, where each element is composed of a triple of time, channel, and value. The encoder and decoder of the Tripletformer are designed with attention layers and fully connected layers, enabling the model to effectively process the presented set elements.
We evaluate the Tripletformer against a range of baselines on multiple real-world and synthetic datasets and show that it produces more accurate and certain interpolations. Results indicate an improvement in negative loglikelihood error by up to 32% on real-world datasets and 85% on synthetic datasets when using the Tripletformer compared to the next best model.

Artificial Intelligence (AI)-based methods in the energy sector challenge companies, organizations, and societies. Organizational issues include traceability, certifiability, explainability, responsibility, and efficiency. Societal challenges include ethical norms, bias, discrimination, privacy, and information security. Explainable Artificial Intelligence (XAI) can address these issues in various application areas of the energy sector, e.g., power generation forecasting, load management, and network security operations. We derive Key Topics (KTs) and Design Requirements (DRs) and develop Design Principles (DPs) for efficient XAI applications through Design Science Research (DSR). We analyze 179 scientific articles to identify our 8 KTs for XAI implementation through text mining and topic modeling. Based on the KTs, we derive 15 DRs and develop 18 DPs. After that, we discuss and evaluate our results and findings through expert surveys. We develop a Three-Forces Model as a framework for implementing efficient XAI solutions. We provide recommendations and a further research agenda.

In this paper, a localization function for an autonomous vehicles (AV) is implemented in a cyber-physical laboratory test field. This function is adapted to the properties of the cyber-physical system and can be used in variety of projects of different domains, such as indoor road traffic simulation as well as in the context of Industry 4.0 applications. The great advantage of this feature is that the position of the AV platform can be determined in parallel by two different methods: a self-localization algorithm using sensors on the vehicle and through the sensor in the environment. These two data sources are transmitted via UDP in the CPS laboratory test field, which is used as the basis for the subsequent application (navigation).