Hazaël Jones

The TARDIS Framework: Intelligent Virtual Agents for Social Coaching in Job Interviews

The TARDIS project aims to build a scenario-based serious-game simulation platform for NEETs and job-inclusion associations that supports social training and coaching in the context of job interviews. This paper presents the general architecture of the TARDIS job interview simulator, and the serious game paradigm that we are developing.

Towards a Socially Adaptive Virtual Agent

This paper presents a socially adaptive virtual agent that can adapt its behaviour according to social constructs (e.g. attitude, relationship) that are updated depending on the behaviour of its interlocutor. We consider the context of job interviews with the virtual agent playing the role of the recruiter. The evaluation of our approach is based on a comparison of the socially adaptive agent to a simple scripted agent and to an emotionally-reactive one. Videos of these three different agents in situation have been created and evaluated by 83 participants. This subjective evaluation shows that the simulation and expression of social attitude is perceived by the users and impacts on the evaluation of the agent’s credibility. We also found that while the emotion expression of the virtual agent has an immediate impact on the user’s experience, the impact of the virtual agent’s attitude expression’s impact is stronger after a few speaking turns.

Who’s Afraid of Job Interviews? Definitely a Question for User Modelling

We define job interviews as a domain of interaction that can be modelled automatically in a serious game for job interview skills training. We present four types of studies: (1) field-based human-to-human job interviews, (2) field-based computer-mediated human-to-human interviews, (3) lab-based wizard of oz studies, (4) field-based human-to-agent studies. Together, these highlight pertinent questions for the user modelling field as it expands its scope to applications for social inclusion. The results of the studies show that the interviewees suppress their emotional behaviours and although our system recognises automatically a subset of those behaviours, the modelling of complex mental states in real-world contexts poses a challenge for the state-of-the-art user modelling technologies. This calls for the need to re-examine both the approach to the implementation of the models and/or of their usage for the target contexts.

A new approach for zoning irregularly-spaced, within-field data

Abstract Management zones can be defined as homogeneous regions for which specific management decisions are to be considered. The delineation of these management units is important because it enables or at least facilitate growers and practitioners performing site specific management. The delineation of management zones has essentially been performed by (i) clustering techniques or (ii) segmentation algorithms arising from the image processing domain. However, the first approach does not take into account the spatial relationships in the data, and is prone to generate a large number of fragmented zones while he second methodology has only been dedicated to regularly-spaced, within-field data. This work proposes a new approach to generate contiguous management zones from irregularly-spaced within-field observations, e.g. within-field yield, soil conductivity, soil samples, which are a very important source of data in precision agriculture studies. A seeded region growing and merging algorithm has been specifically designed for these irregularly-spaced observations. More specifically, a Voronoi tessellation was implemented to define spatial relationships between neighbouring observations. Seeds were automatically placed at specific locations across the fields and management zones were first expanded from these seeds. The merging procedure aimed at generating more manageable and interpretable zones. The merging algorithm was defined in a way that made it possible to incorporate machinery and technical management constraints. Experiments demonstrated that the proposed methodology was able to generate relatively compact and contiguous management zones. Furthermore, machinery and technical constraints were shown to significantly influence the results of the delineation which proved the importance of accounting for these considerations.

A general method to filter out defective spatial observations from yield mapping datasets

Yield maps are recognized as a valuable tool with regard to managing upcoming crop production but can contain a large amount of defective data that might result in misleading decisions. These anomalies must be removed before further processing to ensure the quality of future decisions. This paper proposes a new holistic methodology to filter out defective observations likely to be present in yield datasets. The notion of spatial neighbourhood has been refined to embrace the specific characteristics of such on-the-go vehicle based datasets. Observations are compared with their newly-defined spatial neighbourhood and the most abnormal ones are classified as defective observations based on a density-based clustering algorithm. The approach was conceived to be as non-parametric and automated as far as possible to pre-process a growing number of datasets without supervision. The proposed approach showed promising results on real yield datasets with the detection of well-known sources of errors such as filling and emptying times, speed changes and non-fully used cutting bar.

Fuzzy Rules for Events Perception and Emotions in an Agent Architecture

In complex simulations, multi-agents systems allow to model virtual humans with an explicit cognitive process representation. However, this cognitive process is hard to model and is therefore generally simplified in an application-dependant way. In order to improve the realism of individual and collective behavior of these agents, we propose to integrate the perception of events and the computation of agents emotions in a fuzzy framework. The modeling of the perception and its effect on emotions through fuzzy rules enables the agents to consider properly the virtual environment. We show how different kinds of fuzzy rules can help in the calculus of emotions. Computation of emotions is based on the evaluation of events’ occurrence. Once the events are perceived by the agents, our method uses the desirability of these events to compute emotions relevant to crisis situations. We illustrate this model with a traffic simulation example.

An optimisation-based approach to generate interpretable within-field zones

The paper proposes a numerical criterion to evaluate zoning quality for a given number of classes. The originality of the criterion is to simultaneously quantify how zones are heterogeneous on the whole field under study and how neighbouring zones are similar. This approach allows comparison between maps either with different zones or different labels, which is of importance for zone delineation algorithms aiming at maximizing inter-zone variability. In addition, this study also proposes an optimisation procedure that yields interpretable within-field zones in which each zone is assigned a clear label. The zoning procedure involves contour delineation based on quantile values. The key point of the paper is to use the proposed numerical zoning quality criterion to guide the optimisation procedure showing the complementarity of both proposals in delineating relevant within-field zones. In order to demonstrate the relevancy of the criterion, the zoning procedure and the implementation of both together, the method was tested on 50 theoretical fields with known variability and known spatial structure. A real plot with yield monitoring data was also used to demonstrate the value of the approach on a real case. Results show the relevancy of the methodology to compare maps with different zones and to sort them. Results also demonstrate the interest of the optimisation procedure to provide a ranked set of possible maps with different within-field zones. This set of relevant maps may constitute a decision support for practitioners who may consider additional expert information to choose the most appropriate map in the specific conditions under consideration.

An Affective Model for a Virtual Recruiter in a Job Interview Context

The TARDIS1 project aims to build a scenario-based serious-game simulation platform for young people at risk of exclusion to explore, practice and improve their social skills. This paper presents a model for socio-emotionally realistic virtual agents in the context of job interview simulations.