Purpose instructions and task models in multiple-text reading

Along tasks analysis and modeling history it has been demonstrated by experience that task modeling activities become cumbersome when performed on large, real-life systems. However, one of the main goals of task models is to provide designers with a structured and complete description of the users tasks especially when these user tasks are numerous and/or complex. Several authors proposed to handle that problem by providing tools aiming at supporting both construction and understanding (usually via simulation) of models. One of the most popular examples is CTTE environment which is dedicated to the engineering of CTT task models. The paper shows how to extend notations for task description with two kinds of mechanisms: composition and refinement/abstraction. Refinement/abstraction mechanisms make it possible to decompose a task model into several models and to interconnect them. Composition mechanisms make it possible to define communication means between task models. The paper proposes a precise definition of these mechanisms, their integration into a notation for describing task models and demonstrates that altogether, these two structuring mechanisms support the effective exploitation of task models for large scale application. The use of the mechanisms is presented on a real-life case study from the space domain describing operators' tasks to monitor a satellite and manage failures.

The importance of task analysis and modelling to software system development is well recognised. After all, the central mission for most software systems is to help users accomplish their tasks. However, while object-oriented analysis and design (OOAD) has been established as the major paradigm in the software industry, effectively incorporating task analysis and modelling into that process is still in the realm of research. In this paper, we propose a novel approach for integrating OOAD and Task Modelling. By exploiting the common semantic ground between task models and system behaviour models, we describe how UML diagrams, such as use cases, use case diagrams and scenario diagrams, can be automatically generated from task models represented in a semi-formal notation. We demonstrate our approach with a working example.

This paper aims at clarifying the articulation between the task models and system models encountered in CHI design practices. We demonstrate how the use of a formal task model may enhance the design of interactive systems, by providing quantitative results on which designers may base their decisions. We also demonstrate that it is possible to describe both task and system models within the same formal framework. This enables us firstly to formally prove that task and system models comply with each other, and secondly to perform quantitative analysis on the combination of task and system models. We propose a software design life cycle integrating formal task and system modelling. The approach is illustrated by a toy example which, despite its small size, allows us to develop both task and device models and to perform several iterations on the life cycle. The device and tasks are modelled using the Interactive Cooperative Objects (ICO) formalism, which is based on Petri nets and on the object-oriented approach. The formality of Petri nets allows for axiomatic validation of isolated and interacting subsystems.

Over the last few years, Model-Based User Interface Design has become an important tool for creating multi-device User Interfaces. By providing information about several aspects of the User Interface, such as the task for which it is being built, different User Interfaces can be generated for fulfilling the same needs although they have a different concrete appearance. In the process of making User Interfaces with a Model-Based Design approach, several models can be used: a task model, a dialog model, a user model, a data model,etc. Intuitively, using more models provides more (detailed) information and will create more appropriate User Interfaces. Nevertheless, the designer must take care to keep the different models consistent with respect to each other. This paper presents an algorithm to extract the dialog model (partially) from the task model. A task model and dialog model are closely related because the dialog model defines a sequence of user interactions, an activity chain, to reach the goal postulated in the task specification. We formalise the activity chain as a State Transition Network, and in addition this chain can be partially extracted out of the task specification. The designer benefits of this approach since the task and dialog model are consistent. This approach is useful in automatic User Interface generation where several different dialogs are involved: the transitions between dialogs can be handled smoothly without explicitely implementing them.

This paper addresses the problem of the articulation between task modelling and system modelling in the design of interactive software. We aim at providing solutions allowing the software designers to use efficiently task models during the design process, and to check that the software being built actually corresponds to the requirements elicited during the task analysis phase. The proposed approach is twofold: Firstly, we use the User Action Notation, a semi-formal task modelling formalism, and we present a translation scheme allowing to transform the User Action Notation constructs into Petri nets. Secondly, we use the Interactive Cooperative Objects formalism (based on Petri nets and on the object-oriented approach) to build the model of the system. We finally use the mathematical analysis techniques stemming from the Petri net theory to analyse and validate the cooperation between task models and system model. The approach is presented through a case study, showing the User Action Notation task models, the equivalent Petri net models and the Interactive Cooperative Object system model.

Synergistic modelling of tasks, users and systems using formal specification techniques

Models allow us to describe complex systems at different abstract and conceptual levels, hence amplify our analytical and problem solving capabilities, However, a lot of human effort and experience is needed to build correct models, and to translate them to concrete artifacts: in our case a usable user interface. This paper introduces the concept of task and pattern models to leverage the process of task modeling, and show how it can help build generic task models, link them, and instantiate them more readily. Once seen as patterns, we will demonstrate that task models can be disseminated and reused more easily by representing them as predefined types.

Natural language explanation (NLE) models aim at explaining the decision-making process of a black box system via generating natural language sentences which are human-friendly, high-level and fine-grained. Current NLE models <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> Throughout this paper, we refer to NLE models as Natural Language Explanation models aimed for vision and vision-language tasks. explain the decision-making process of a vision or vision-language model (a.k.a., task model), e.g., a VQA model, via a language model (a.k.a., explanation model), e.g., GPT. Other than the additional memory resources and inference time required by the task model, the task and explanation models are completely independent, which disassociates the explanation from the reasoning process made to predict the answer. We introduce NLX-GPT, a general, compact and faithful language model that can simultaneously predict an answer and explain it. We first conduct pre-training on large scale data of image-caption pairs for general understanding of images, and then formulate the answer as a text prediction task along with the explanation. Without region proposals nor a task model, our resulting overall framework attains better evaluation scores, contains much less parameters and is 15× faster than the current SoA model. We then address the problem of evaluating the explanations which can be in many times generic, data-biased and can come in several forms. We therefore design 2 new evaluation measures: (1) explain-predict and (2) retrieval-based attack, a selfevaluation framework that requires no labels. Code is at: https://github.com/fawazsammani/nlxgpt.

Specifying a dialog model from a task model is a frequent process in model-based approaches. However, as several task model formalisms and notations exist, specifying such a dialog model may depend on the task model used for specification. To relax this constraint and to produce usable dialog in the models, a first set of generic guidelines for specifying dialogs is introduced. These guidelines are considered generic in the sense that they can be applied for several types of task and dialog models. These guidelines are more pragmatic as they are primarily based on the temporal relations specified in the task model.

Task models are widely used in the field of user interface development. They represent a human actor’s performance or the co-operation of a group of people on or together with a system. For considerable time, it was an open problem in the field how to switch from the analyzing step of task analysis and modeling to the synthesizing step of user interface design. In the meantime, interesting approaches have shown up dealing with this problem and helping to bridge the gap between task modeling and user interface development. In this chapter, some of these approaches are discussed, together with recent concepts used to improve the usability of user interfaces based upon underlying task models.

The success of model-based approaches to user interface design depends on the ability to solve the mapping problem as well as on the availability of tools able to reduce the effort of establishing and maintaining of links between models throughout the development life cycle. In this paper a tool supporting a small set of mapping rules is presented. The tool enables the designer to produce task model fragments at operational level based on the patterns of mapping between task and domain models. The task model fragments are generated in XML format and can be further loaded in task modeling tools like CTTE or Teresa.

Task modelling and user interface design are iterative development processes. In this paper we will take a closer look to the relationship between three models: task, application domain and presentation. The objective is to explore heuristics aiming at progressively derive the presentation model from application-domain and task model.

A user-centred approach to systems design and evaluation is presented. In particular task modelling is proposed as a bridging technology for understanding users’ needs, feeding this understanding into design and evaluating the usability of design choices. The approach is proposed as a general methodological framework for the design, the development and the evaluation of interactive systems. We discuss the beneficial bridging effects of task modelling throughout the development life cycle presenting the experience of the ESPRIT Project CHARADE, an integrated system for handling environmental emergencies. The main aspects of our experience are reported highlighting how the system life cycle has been characterized by an early and continual involvement of the final user, and how task analysis and modelling are shared resources for the entire development process.

Mixed Interactive Systems (MIS) is a generic term encompassing mixed and augmented reality, augmented virtuality, and tangible interfaces systems. The work we present in this paper deals with the design of such systems. Several models have been proposed to describe mixed interactive systems. However, these models neither integrate MIS in the global user activity, nor take into account the dynamics of mixed interactive situations. Task models offer an interesting approach to cover these aspects. The purpose of our work is to contribute to the design of MIS by articulating mixed interaction models and tasks models. This paper presents complementarities and common points between these models, which will be used as the basis for the rules to articulate the task and interaction models.

The model-based approach to user interface design relies on developing separate models capturing various aspects about users, tasks, application domain, presentation and dialog structures. This paper presents a task modeling approach for user interface design and aims at exploring mappings between task, domain and presentation models. The basic idea of our approach is to identify typical configurations in task and domain models and to investigate how they relate each other. A special emphasis is put on applicationspecific functions and mappings between domain objects and operational task structures. In this respect, we will address two layers in task decomposition: a functional (planning) layer and an operational layer.