Philippe Caillou

A Simple-to-Use BDI Architecture for Agent-Based Modeling and Simulation

With the increase of computing power and the development of user-friendly multi-agent simulation frameworks, social simulations have become increasingly realistic. However, most agent architectures in these simulations use simple reactive models. Cognitive architectures face two main obstacles: their complexity for the field-expert modeler, and their computational cost. In this paper, we propose a new cognitive agent architecture based on the Belief-Desire-Intention paradigm integrated into the GAMA modeling platform. Based on the GAML modeling language, this architecture was designed to be simple-to-use for modelers, flexible enough to manage complex behaviors, and with low computational cost. This architecture is illustrated with a simulation of the evolution of land-use in the Mekong Delta.

Refounding of the activity concept? Towards a federative paradigm for modeling and simulation

Currently, the widely used notion of activity is increasingly present in computer science. However, because this notion is used in specific contexts, it becomes vague. Here, the notion of activity is scrutinized in various contexts and, accordingly, put in perspective. It is discussed through four scientific disciplines: computer science, biology, economics, and epistemology. The definition of activity usually used in simulation is extended to new qualitative and quantitative definitions. In computer science, biology and economics disciplines, the new simulation activity definition is first applied critically. Then, activity is discussed generally. In epistemology, activity is discussed, in a prospective way, as a possible framework in models of human beliefs and knowledge.

GAMA: A Spatially Explicit, Multi-level, Agent-Based Modeling and Simulation Platform

Agent-based modeling is now widely used to investigate complex systems but still lacks integrated and generic tools to support the representation of features usually associated with real complex systems, namely rich, dynamic and realistic environments or multiple levels of agency. The GAMA platform has been developed to address such issues and allow modelers, thanks to the use of a high-level modeling language, to build, couple and reuse complex models combining various agent architectures, environment representations and levels of abstraction.

A BDI Agent Architecture for the GAMA Modeling and Simulation Platform

With the increase of computing power and the development of user-friendly multi-agent simulation frameworks, social simulations have become increasingly realistic. However, most agent architectures in these simulations use simple reactive models. Indeed, cognitive agent architectures face two main obstacles: their complexity for the field-expert modeler, and their computational cost. In this paper, we propose a new cognitive agent architecture based on the BDI (Belief-Desire-Intention) paradigm integrated into the GAMA modeling platform and its GAML modeling language. This architecture was designed to be simple-to-use for modelers, flexible enough to manage complex behaviors, and with low computational cost. An experiment carried out with different profiles of end-users shows that the architecture is actually usable even by modelers who have little knowledge in programming and in Artificial Intelligence.

Automated multi-agent simulation generation and validation

Multi-agent based simulation (MABS) is increasingly used for social science studies. However, few methodologies and tools exist. A strong issue is the choice of the number of simulation runs and the validation of the results by statistical methods. In this article, we propose a model of tool which automatically generates and runs new simulations until the results are statistically valid using a chi-square test. The choice of the test configuration allows both a general overview of the variable links and a more specific independence analysis. We present a generic tool for any RePast-based simulation and apply it on an Academic Labor Market economic simulation.

Pride and Prejudice on a Centralized Academic Labor Market

The Academic Labor Market in France can be viewed as a constrained Stable Marriage problem, pairing universities and candidates according to their (elitist) preferences. A Multi-Agent based model, calibrated after the empirical evidence, is used to investigate how universities can recruit the best candidates with high confidence. Extensive simulations suggest that universities can be divided in four categories: top and medium universities have no difficulty in attracting the candidates they have selected, contrarily to good and bad universities. In this paper, a learning mechanism is presented: universities are allowed to tune their expectations depending on whether they did succeed to attract candidates in the previous recruitment rounds. The impact of over/under estimations is analyzed with respect to the hiring efficiency and quality.

Intelligent Digital Learning: Agent-Based Recommender System

In the context of intelligent digital learning, we propose an agent-based recommender system that aims to help learners overcome their gaps by suggesting relevant learning resources. The main idea is to provide them with appropriate support in order to make their learning experience more effective. To this end we design an agent-based cooperative system where autonomous agents are able to update recommendation data and to improve the recommender outcome on behalf of their past experiences in the learning platform.

Multidimensional decision model for classifying learners: the case of massive online open courses (MOOCs)

Abstract A new era of learning is arising, due to the development of the digital world and the maturity of web technologies. Massive open online courses (MOOCs) have emerged in this context and are challenging classical learning in spite of boundaries of time and space. They aim to provide good-quality education to masses that cannot be part of traditional university and school learning processes. This emancipation is coupled with a large amount of data collection via learning platforms. These data constitute a great opportunity to study interactions and to profit from this, in order to optimise learning and knowledge transfer. The challenge in creating value is how to represent, analyse and reuse data in order to characterise learners and so lead to better knowledge transfer. We focus, in this article, on the context of MOOC platforms based on Open edX (The leading online courses platform, initially developed by MIT and Harvard). The purpose of this paper is to propose a decision model that takes advantage of the different data-sets available on the learning platform in order to classify learners based on their behaviour and expertise. To this end, we propose a multi-agent approach that represents human actors using intelligent agents. A coordinator agent implements the multi-criteria decision model in order to optimise the learning process. This agent employs clustering algorithms and has an overview of the learning platform that enables it to assist learners and to learn from experiences and about other agents’ behaviour.

Simulation of the Rungis Wholesale Market: Lessons on the Calibration, Validation and Usage of a Cognitive Agent-Based Simulation

We present some methodological lessons and thoughts inferred from a research we are making on a simulation of the Rungis Wholesale Market (in France) using cognitive agents. The implication of using cognitive agents with an objective of realism at the individual level question some of the classical methodological assertions about simulations. Three such lessons are of particular interest: the calibration and validation focus on individuals rather than global values (1); the definition of the simulation model is made independently from the research objectives (2), and without targeting the usual objective of hypothesis simplicity (3).

Economy-Driven Shaping of Social Networks and Emerging Class Behaviors

Agent-based Computational Economics (ACE) is a powerful framework for studying emergent complex systems resulting from the interactions of agents either mildly rational, or with incomplete information (Axelrod, 2004; Tesfatsion, 2002, 2006) or driven by the social network (Bala and Goyal, 2003; Carayol and Roux, 2004; Slikker and van den Nouweland, 2000).