Sybille Caffiau

Evaluation of a Context-Aware Voice Interface for Ambient Assisted Living: Qualitative User Study vs. Quantitative System Evaluation

This article presents an experiment with seniors and people with visual impairment in a voice-controlled smart home using the Sweet-Home system. The experiment shows some weaknesses in automatic speech recognition that must be addressed, as well as the need for better adaptation to the user and the environment. Users were disturbed by the rigid structure of the grammar and were eager to adapt it to their own preferences. Surprisingly, while no humanoid aspect was introduced in the system, the senior participants were inclined to embody the system. Despite these aspects to improve, the system has been favorably assessed as diminishing most participant fears related to the loss of autonomy.

Building and Using Home Automation Systems: A Field Study

These last years, several new home automation boxes appeared on the market, the new radio-based protocols facilitating their deployment with respect to previously wired solutions. Coupled with the wider availability of connected objects, these protocols have allowed new users to set up home automation systems by themselves. In this paper, we relate an in situ observational study of these builders in order to understand why and how the smart habitats were developed and used. We led 10 semi-structured interviews in households composed of at least 2 adults and equipped for at least 1 year, and 47 home automation builders answered an online questionnaire at the end of the study. Our study confirms, specifies and exhibits additional insights about usages and means of end-user development in the context of home automation.

Assessment of Object Use for Task Modeling

Past research in task modeling suggests the need to introduce objects when using task models for the design of interactive applications. Objects are however rarely included in the task model notations and formalisms. Furthermore, when part of the formalism, their definition is usually informal; and the supporting tool does not generally take them into account for simulation. K-MADe is the first tool that fully uses objects for condition evaluations during task model simulation. This paper presents an evaluation investigating the usage of formal objects with K-MADe. The results show that whilst object concepts seem to be essential in the task model process, their usage and manipulation is not easy.

Generating interactive applications from task models: a hard challenge

Since early ergonomics, notations have been created focusing on the activities, jobs and task descriptions. However, the development of a wide variety of devices led to the generation of different interfaces from the same description of the tasks. The generation of complete current interfaces needs different types of information, some of which are not represented in usual task models. The goal of this paper is to present information that seems to be lacking in the task models.

Hierarchical Structure: A Step for Jointly Designing Interactive Software Dialog and Task Model

In order to design interactive applications, the first step is usually the definition of user needs. While performing this step, activities may be modeled using task models. Some task model components express scheduling information that describes the task dynamics. According to a model-based approach, the dynamics of applications (i.e.: the dialog) can be formalized using a dialog model. Several approaches seek to exploit the task model information to perform the dialog model. This paper aims to show that the use of the hierarchical dialog model facilitates its design according to task model information during the whole iterative design process.

A Task Based model for Récit Generation from Sensor Data: An Early Experiment

Automatic story generation is the subject of a growing research effort. However, in this domain, stories are generally produced from fictional data. In this paper, we present a task model used for automatic story generation from real data focusing on the narrative planning. The aim is to generate recits (stories) from sensors data acquired during a ski sortie. The model and some preliminary analysis are presented which suggest the interest of the approach.

The Sweet-Home project: Audio processing and decision making in smart home to improve well-being and reliance

The Sweet-Home project aims at providing audio-based interaction technology that lets the user have full control over their home environment, at detecting distress situations and at easing the social inclusion of the elderly and frail population. This paper presents an overview of the project focusing on the implemented techniques for speech and sound recognition as context-aware decision making with uncertainty. A user experiment in a smart home demonstrates the interest of this audio-based technology.

Vérification de cohérence entre modèles de tâches et de dialogue en conception centrée-utilisateur

Model-based approach usage is increasing in interactive application design. Models are used to express the viewpoints of every design participant (particularly the user). This paper presents an approach for checking consistency between task models and dialog models during iterative application design. At the end, an example of the proposed approach is presented. It is based on the K-MAD task model, the IH dialog model (hierarchical interactors) and the use of Praxis (on witch rules are expressed).

Formally expressing the users' objects world in task models

While past research presents objects as essential entities in task modeling, they are in fact rarely used. In the first tool that truly considered objects, K-MADe, two main reasons may explain this limited use: an incomplete module of object description (in the tool, the expression coverage of the object concept is not wide enough) and the usability problems of K-MADes specific interface. This paper presents a study on the object expression coverage. From case studies, we identify limitations and we infer modification on the K-MADe object entity definitions. These modifications aim to increase the expressive power of objects.

Natural language generation to support the understanding of task models: a preliminary study

During the design of IS, several stakeholders, each with their own expertise and culture, will use the task model (TM). Then, TM must be understood by several kinds of stakeholders. Adapting the presentation of the TM information to the stakeholder and the use he or she makes of it, is therefore a key point for them to be understood. however, none of the tools for use information of the TM allow to understand the modeled activity without knowing the notation used. In this paper, we propose to use a presentation in the textual form as a tool for verification and consultation of the task model. Two evaluations were conducted on the use of this presentation for (1) the correction of a model by its designers and (2) the understanding of a task model by computer scientists who did not design the model. These evaluations show that the textual presentation of the task model provides a complementary point of view allowing a better understanding and facilitating the improvement of the model.