Human Data Interaction

Tristan Henderson, Richard Mortier, Hamed Haddadi, Jon Crowcroft, Derek McAuley


Recent years have seen increasing collection and use of personal data – that is, data about us and data produced by us – both public and private, about us and our activities. Such data include purchasing habits (on- and off-line), financial data, communications data (from phone call records to social media content), and more. There has been similar growth in applications that provide us with benefits from our own publicly-released data: traffic reports on Google Maps, crowd-sourced road conditions on Waze, and optimised bus routes with mobile phone data. The impact of this data processing is pervasive and wide ranging – it informs credit ratings, online advertising, retailing, and is used for a variety of other predictions and inferences, from sexual orientation to voting preferences. These data are at the heart of many Internet business models, particularly those based on advertising and market intelligence.

We deliberately adopt the phrase HDI by analogy with HCI, but the two can be clearly distinguished. Unlike previous definitions of Human-Data Interaction focused on visualisation, primarily embodied, of large datasets, we believe that HDI concerns interaction generally between humans, datasets and analytics, but not the general study of interaction with computer systems that is HCI. HDI refers to the analysis of the individual and collective decisions we make and the actions we take, whether as users of online systems or as subjects of data collection. The term makes explicit the link between individuals and the signals they emit as data (e.g., location, shopping trends, search terms), as the richness, pervasiveness and impact of these models and techniques continues to grow.

There are two features that make HDI interesting. First, as recent experience with online social networks and the NSA’s PRISM have shown, the impact of the inferences drawn from public personal data can affect the market value of billion dollar corporations or move the use of national infrastructure outside expected parameters. Second, inferences drawn from on- and off-line private personal data, such as passive measurement, location, and communications, create virtual personalities for each individual. Thus HDI contains a simultaneous mix of two contrasting features: sheer scale and personal richness.


Mortier, R., Haddadi, H., Henderson, T., McAuley, D. & Crowcroft, J., ‘ Challenges & Opportunities in Human-Data Interaction ‘, 2013, Proceedings of DE2013: Open Digital — The Fourth Annual Digital Economy All Hands Meeting.

Mortier, R., Haddadi, H., Henderson, T., McAuley, D. & Crowcroft, J., ‘ Human-Data Interaction: The Human Face of the Data-Driven Society ‘, 2015