< Back to previous page

Project

Towards more Transparent Human-Data Interaction: Designing for Uncertainty in Digital Humanities Visualization

While the increased datafication of society brings about obvious benefits, it is also facing criticism for its treatment of data as an absolute, objective and raw representation of truth. To mitigate this critique, data visualisation research is developing effective ways to convey the uncertainty that underlies it.

Although data uncertainty is commonly associated with statistical dimensions such as modelling errors or forecasting probabilities, recent empirical research has demonstrated that other types of uncertainties exist when people work with data. This dissertation focuses on indirect qualitative uncertainty, that is prevalent in humanistic research because of the partial and incomplete nature of humanities data. Unlike its quantifiable counterpart, indirect uncertainty is mostly handled by relying on tacit knowledge, which consequently makes it challenging to capture and then communicate to others.

This dissertation aims to support digital humanists to more confidently engage with data visualisation by communicating  uncertainty and by encouraging more participatory, interdisciplinary ways of working. Accordingly it (1) examines how indirect, qualitative uncertainty manifests in the research context of the digital humanities, (2) investigates how to communicate it in digital humanities visualisations and (3) develops activities to make the generation of digital humanities visualization more participatory. 

The empirical contributions of this dissertation are grounded in the Sagalassos Archaeological Research Project, a long-standing interdisciplinary project that includes multiple scientists from the fields of archaeology, ecology, human geography and urban planners among others. Drawing from human-computer interaction research methods, the first part of the thesis synthesises three cases in which different origins of uncertainty are examined:

- Case 1 analyses existing practise for handling qualitative uncertainty in digital humanities visualisation.
- Case 2 examines how to communicate data frictions such as methodological inconsistency when synthesising data from an interdisciplinary group of scientists.
- Case 3 investigates how to represent implicit errors such as human subjectivity when visualising archaeological research data.

The methodological contributions of this dissertation include activities that make the design process of digital humanities visualizations more participatory. This second part of the thesis proposes four participatory techniques:

- GoCo uses storytelling and tangible tokens to help an interdisciplinary group of experts articulate their data and align conflicting points-of-view,
- Co-gnito examines how to elicit and represent urban mental maps of people on a single physicalisation.
- Data Badges examines how to elicit personal data characteristics through a DIY physicalisation toolkit meant for conference participants.

By synthesising the insights from both parts, this dissertation brings forward a holistic understanding of uncertainty in digital humanities visualisation which includes the data generation and focuses on experience rather than quantification. It thus contributes to the design of visualisations that enable transparent and inclusive discussions around data, which are a necessary step for establishing more confident, open and reflexive science.

Date:22 Jan 2018 →  13 May 2022
Keywords:visual analytics, visualization, design, archaeology, data visualisation, data uncertainty
Disciplines:Architectural engineering, Art studies and sciences
Project type:PhD project