EuroVisPosters
Permanent URI for this community
Browse
Browsing EuroVisPosters by Author "b7736374-f807-4f21-82f0-e4a978fec7d3"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item A Design Space for Explainable Ranking and Ranking Models(The Eurographics Association, 2022) Hazwan, Ibrahim Al; Schmid, Jenny; Sachdeva, Madhav; Bernard, Jürgen; Krone, Michael; Lenti, Simone; Schmidt, JohannaItem ranking systems support users in multi-criteria decision-making tasks. Users need to trust rankings and ranking algorithms to reflect user preferences nicely while avoiding systematic errors and biases. However, today only few approaches help end users, model developers, and analysts to explain rankings. We report on the study of explanation approaches from the perspectives of recommender systems, explainable AI, and visualization research and propose the first cross-domain design space for explainers of item rankings. In addition, we leverage the descriptive power of the design space to characterize a) existing explainers and b) three main user groups involved in ranking explanation tasks. The generative power of the design space is a means for future designers and developers to create more target-oriented solutions in this only weakly exploited space.Item Interaction Tasks for Explainable Recommender Systems(The Eurographics Association, 2023) Al-Hazwani, Ibrahim; Alahmadi, Turki; Wardatzky, Kathrin; Inel, Oana; El-Assady, Mennatallah; Bernard, Jürgen; Gillmann, Christina; Krone, Michael; Lenti, SimoneIn the modern web experience, users interact with various types of recommender systems. In this literature study, we investigate the role of interaction in explainable recommender systems using 27 relevant papers from recommender systems, humancomputer interaction, and visualization fields. We structure interaction approaches into 1) the task, 2) the interaction intent, 3) the interaction technique, and 4) the interaction effect on explainable recommender systems. We present a preliminary interaction taxonomy for designers and developers to improve the interaction design of explainable recommender systems. Findings based on exploiting the descriptive power of the taxonomy emphasize the importance of interaction in creating effective and user-friendly explainable recommender systems.Item ORD-Xplore: Bridging Open Research Data Collections through Modality Abstractions(The Eurographics Association, 2023) Sachdeva, Madhav; Blum, Michael; Stricker, Yann; Schreck, Tobias; Mumenthaler, Rudolf; Bernard, Jürgen; Gillmann, Christina; Krone, Michael; Lenti, SimoneWe present ORD-Xplore, an approach to bridge gaps between digital editions, which represent valuable collections of multiple digitized research artifacts. However, digital editions often co-exist isolated, making it difficult for researchers to access, find, and re-use open research data from multiple digital editions. An ultimate goal is to unify library services across editions, even for editions with heterogeneity. In ORD-Xplore, we utilize abstraction methods from visualization research to help digital librarians identify unifying data modalities, as one important step towards standardization of heterogeneous digital editions.