In this talk, I will explore different perspectives on Human-Data Interaction – the study of how people engage with, experience, and make sense of data. My research aims to increase transparency in the decisions that shape data – from the questions we ask, the ways we collect and document data to how we analyze, communicate, and reuse data. A key challenge in this space is explaining both data and models to diverse audiences and developing tools that support these efforts.

I will present two related areas of research that investigate how people interact with data and models and how they understand them. The first area focuses on data-centric sensemaking and touches upon the concept of "data needs'', and the notion of data as a dynamic process rather than a static artifact. The second area explores how different audiences interpret data through visualizations, with particular attention to the role of messages attached to visual data representations. Visualizations are everyday tools for communicating data, however, it is not clear whether audiences interpret them as their creators intended.

Drawing on my ongoing project, Talking Charts, I will illustrate these ideas with examples from research engaging both expert and general audiences across various visualization contexts. Through this, I aim to invite reflection on how we can design more meaningful data experiences.

Data integration services are of utmost importance in research data infrastructures like the NFDI, enabling the harmonization of diverse data sources. While significant progress has been made in automatic schema mapping and protocol development, non-functional requirements play a vital role in ensuring fair and transparent collaboration between data infrastructures and providers. Key non-functional requirements include access control, resource usage limitations, quota tracking, and usage statistics reporting.

These are crucial for maintaining provider trust and data integrity. For instance, access control ensures that only authorized users can access sensitive data, while usage statistics allow providers to track the impact of their contributions. Balancing these requirements with user experience presents a challenge. End-users expect scalable and fast data access without noticeable restrictions, necessitating a solution that addresses both provider needs and user expectations.

To address this challenge, we present the XCIP (eXchange Component for Infrastructure Providers), a novel exchange component designed to facilitate connection and communication between data infrastructures and providers, which builds upon our previous work, the SDExL (Spatiotemporal Data Exchange Layer). XCIP not only addresses non-functional requirements but also introduces innovative mechanisms for data caching and access optimization. These features enhance data accessibility for end-users while maintaining transparency, effectively bridging the gap between provider requirements and user expectations. This paper details the design of XCIP, demonstrating how it addresses non-functional requirements while simultaneously improving the end-user experience through transparent data caching and statistics collection.