Contributions in Merlin

The Semantic Web is distributed yet interoperable: Distributed since resources are created and published by a variety of producers, tailored to their specific needs and knowledge; Interoperable as entities are linked across resources, allowing to use resources from different providers in concord. Complementary to the explicit usage of Semantic Web resources, embedding methods made them applicable to machine learning tasks. Subsequently, embedding models for numerous tasks and structures have been developed, and embedding spaces for various resources have been published. The ecosystem of embedding spaces is distributed but not interoperable: Entity embeddings are not readily comparable across different spaces. To parallel the Web of Data with a Web of Embeddings, we must thus integrate available embedding spaces into a uniform space. Current integration approaches are limited to two spaces and presume that both of them were embedded with the same method — both assumptions are unlikely to hold in the context of a Web of Embeddings. In this paper, we present FedCoder— an approach that integrates multiple embedding spaces via a latent space. We assert that linked entities have a similar representation in the latent space so that entities become comparable across embedding spaces. FedCoder employs an autoencoder to learn this latent space from linked as well as non-linked entities. Our experiments show that FedCoder substantially outperforms state-of-the-art approaches when faced with different embedding models, that it scales better than previous methods in the number of embedding spaces, and that it improves with more graphs being integrated whilst performing comparably with current approaches that assumed joint learning of the embeddings and were, usually, limited to two sources. Our results demonstrate that FedCoder is well adapted to integrate the distributed, diverse, and large ecosystem of embeddings spaces into an interoperable Web of Embeddings.

null

Empirical studies show that first- and second-generation immigrants are less likely to be members of sports clubs than their non-immigrant peers. Common explanations are cultural differences and socioeconomic disadvantages. However, lower participation rates in amateur sport could be at least partly due to ethnic discrimination. Are minority ethnic groups granted the same right to belong as their non-immigrant peers? To answer this question, this paper uses publicly available data from a field experiment in which mock applications were sent out to over 1,600 football clubs in Germany. Having a foreign-sounding name significantly reduces the likelihood of being invited to participate. The paper concludes that amateur football clubs are not as permeable as they are often perceived to be. It claims that traditional explanations for lower participation rates of immigrants need to be revisited.

null

We examine how six search engines filter and rank information in relation to the queries on the U.S. 2020 presidential primary elections under the default—that is nonpersonalized—conditions. For that, we utilize an algorithmic auditing methodology that uses virtual agents to conduct large-scale analysis of algorithmic information curation in a controlled environment. Specifically, we look at the text search results for “us elections,” “donald trump,” “joe biden,” “bernie sanders” queries on Google, Baidu, Bing, DuckDuckGo, Yahoo, and Yandex, during the 2020 primaries. Our findings indicate substantial differences in the search results between search engines and multiple discrepancies within the results generated for different agents using the same search engine. It highlights that whether users see certain information is decided by chance due to the inherent randomization of search results. We also find that some search engines prioritize different categories of information sources with respect to specific candidates. These observations demonstrate that algorithmic curation of political information can create information inequalities between the search engine users even under nonpersonalized conditions. Such inequalities are particularly troubling considering that search results are highly trusted by the public and can shift the opinions of undecided voters as demonstrated by previous research.

Inspired by an industry initiative to address the celebrated market for lemons (poor-quality used cars), we investigate how incentives for a permissioned blockchain-based system in the automobile ecosystem can be designed to ensure high-quality data storage and use by different stakeholders. The peer-to-peer distributed ledger platform connects organizations and car owners with disparate interests and hidden intentions. While previous literature has chiefly examined incentives for permissionless platforms, we leverage studies about crowdsensing applications to stimulate research on incentives in permissioned blockchains. This paper uses the action design research approach to create an incentive system featuring a rating mechanism influenced by data correction measures. Furthermore, we propose relying on certain institutions capable of assessing data generated within the system. This combined approach of a decentralized data correction and an institutionalized data assessment is distinct from similar incentive systems suggested by literature. By using an agent-based model with strategy evolution, we evaluate the proposed incentive system. Our findings indicate that a rating-based revenue distribution leads to markedly higher data quality in the system. Additionally, the incentive system reveals hidden information of the agents and alleviates agency problems, contributing to an understanding of incentive design in inter-organizational blockchain-based data platforms. Furthermore, we explore incentive design in permissioned blockchains and discuss its latest implications.

null

We assess how investors’ willingness-to-pay (WTP) for sustainable investments responds to the social impact of those investments, using a framed field experiment. While investors have a substantial WTP for sustainable investments, they do not pay significantly more for more impact. This also holds for dedicated impact investors. When investors compare several sustainable investments, their WTP responds to relative, but not to absolute, levels of impact. Regardless of investments' impact, investors experience positive emotions when choosing sustainable investments. Our findings suggest that the WTP for sustainable investments is primarily driven by an emotional, rather than a calculative, valuation of impact.

null

null