Skip to Main content Skip to Navigation

Designing Representations for Digital Documents

Han Han 1, 2 
2 EX-SITU - Extreme Situated Interaction
Inria Saclay - Ile de France, LISN - Laboratoire Interdisciplinaire des Sciences du Numérique, IaH - Interaction avec l'Humain
Abstract : Millions of users work with documents for their everyday tasks but their user interfaces have not fundamentally changed since they were first designed in the late seventies. Today’s computers come in many forms and are used by a wide variety of users for a wide range of tasks, challenging the limits of current document interfaces. I argue that by focusing on extreme users and taking on a principled perspective, we can design effective and flexible representations to support document-related knowledge work. I first study one of the most common document tasks, text editing, in the context of technical documents. By focusing on legal professionals, one example of extreme document users, we reveal the limits of current word processors. Legal professionals must rely on their memory to manage dependencies and maintain consistent vocabulary within their technical documents. To address these issues, we introduce Textlets, interactive objects that reify text selections into persistent items. We present a proof-of-concept prototype demonstrating several use cases, including selective search and replace, word count, and alternative wording. The observational evaluation shows the usefulness and effectiveness of textlets, providing evidence of the validity of the textlet concept. During my work with legal professionals in the first project, I was introduced to the domain of patent writing and filling. In the patent process, patent attorneys write patent submissions that describe the invention created by the inventor. Patent examiners review the submission and decide whether the submission can be granted as a patent. In collaboration with a European Patent Office, I studied the patent examiners’ search and review process. The study reveals the need to manage text from multiple documents across various interconnected activities, including searching, collecting, annotating, organizing, writing and reviewing, while manually tracking their provenance. I extend Textlets to create Passages, text selection objects that can be manipulated, reused, and shared across multiple tools. Two user studies show that Passages facilitate knowledge workers practices and enable greater reuse of information. These two projects led to another important aspect of knowledge work: file management. I focus on scientists, another example of extreme knowledge workers, to study their document management practices. In an age where heterogeneous data science workflows are the norm, instead of relying on more self-contained environments such as Jupyter Notebooks, scientists work across many diverse tools. They have difficulties using the file system to keep track of, re-find and maintain consistency among related but distributed information. We created FileWeaver, a system that automatically detects dependencies among files without explicit user action, tracks their history, and lets users interact directly with the graphs representing these dependencies and version history. By making dependencies among files explicit and visible, FileWeaver facilitates the automation of workflows by scientists and other users who rely on the file system to manage their data. These three document representations rely on the same underlying theoretical principles: reification, polymorphism and reuse. I reflect on my experience designing and evaluating these representations and propose three new design principles: granularity, individuality and synchronization. Together with the empirical findings from three examples of extreme users, technological demonstration of three proof-of-concept prototypes and three design principles, this thesis demonstrates fresh new approaches to working with documents, a fundamental representation in GUIs. I argue that we should not accept current desktop interfaces as given, and that by taking on a principled and theory-driven perspective we can contribute innovative interface concepts.
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Monday, May 9, 2022 - 11:04:22 AM
Last modification on : Friday, August 5, 2022 - 9:27:34 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03662229, version 1


Han Han. Designing Representations for Digital Documents. Human-Computer Interaction [cs.HC]. Université Paris-Saclay, 2022. English. ⟨NNT : 2022UPASG025⟩. ⟨tel-03662229⟩



Record views


Files downloads