Computer aided multi-agent system engineering

A computer-aided education environment not only extends education opportunities beyond the traditional classroom, but it also provides opportunities for intelligent interface based on agent-based technologies to better support teaching and learning within traditional classrooms. Advances in information technology, such as the Internet and multimedia technology, have dramatically enhanced the way that information and knowledge are represented and delivered to students. The application of agent-based technologies to education can be grouped into two primary categories, both of which are highly interactive interfaces: (1) intelligent tutoring systems (ITS) and (2) interactive learning environments (ILE) (McArthur, Lewis, & Bishay, 1993). Current research in this area has looked at the integration of agent technology into education systems. However, most agent-based education systems under utilize intelligent features of agents such as reactivity, pro-activeness, social ability (Wooldridge & Jennings, 1995) and machine learning capabilities. Moreover, most current agent-based education systems are simply a group of non-collaborative (i.e., non-interacting) individual agents. Finally, most of these systems do not peruse the multi-agent intelligence to enhance the quality of service in terms of content provided by the interfaces. A multi-agent system is a group of agents where agents interact and cooperate to accomplish a task, thereby satisfying goals of the system design (Weiss, 1999). A group of agents that do not interact and do not peruse the information obtained from such interactions to help them make better decisions is simply a group of independent agents, not a multi-agent system. To illustrate this point, consider an ITS that has been interacting with a particular group of students and has been collecting data about these students. Next, consider another ITS which is invoked to deal with a similar group of students. If the second ITS could interact with the first ITS to obtain its data, then the second ITS would be able to handle its students more effectively, and together the two agents would comprise a multi-agent system. Most ITS or ILE systems in the literature do not utilize the power of a multi-agent system. The Intelligent Multi-agent Infrastructure for Distributed Systems in Education (I-MINDS) is an exception. It is comprised of a multi-agent system (MAS) infrastructure that supports different high-performance distributed applications on heterogeneous systems to create a computer-aided, collaborative learning and teaching environment. In our current I-MINDS system, there are two types of agents: teacher agents and student agents. A teacher agent generally helps the instructor manage the real-time classroom. In I-MINDS, the teacher agent is unique in that it provides an automated ranking of questions from the students. This innovation presents ranked questions to the classroom instructor and keeps track of a profile of each class participant reflecting how they respond to the class lectures. A student agent supports a class participant’s real-time classroom experience. In I-MINDS, student agents innovatively support the buddy group formation. A class participant’s buddy group is his or her support group. The buddy group is a group of actual students that every student has access to during real-time classroom activities and with which they may discuss problems. Each of these agents has its interface which, on one hand, interacts with the user and, on the other hand, receives information from other agents and presents those to the user in a timely fashion. In the following, we first present some background on the design choice of I-MINDS. Second, we describe the design and implementation of I-MINDS in greater detail, illustrating with concrete examples. We finalize with a discussion of future trends and some conclusions drawn from the current design.

Download Full-text

An ontology-based computer-aided diagnosis system in African traditional medicine

Kybernetes ◽

10.1108/k-02-2015-0053 ◽

2016 ◽

Vol 45 (1) ◽

pp. 30-50 ◽

Cited By ~ 3

Author(s):

Armel Ayimdji Tekemetieu ◽

Souleymane KOUSSOUBE ◽

Laure Pauline FOTSO

Keyword(s):

Traditional Medicine ◽

Computer Aided Diagnosis ◽

Multi Agent System ◽

Agent System ◽

Diagnosis System ◽

Content Type ◽

African Traditional Medicine ◽

Computer Aided ◽

Multi Agent ◽

Aided Diagnosis

Purpose – The purpose of this paper is to describe an AI (Artificial Intelligence) that can “think like an African traditional doctor”. The system proposes to model and to use attitudes taken and concepts used by African traditional doctors when facing cases. It is designed to go deep into the concepts of African traditional medicine (ATM) by dealing with all the possible interpretations of those concepts, and to produce more much satisfying and accurate support for medical diagnosis and prescription than existing systems. Design/methodology/approach – To take into account the sometimes strange concepts used and attitudes taken by African traditional healers, including mystical considerations, the system relies on a deep ontology describing all those concepts and attitudes in a more computer readable manner allowing a multi-agent system to have full access to ATM knowledge. Ethnological inquiries, literary analysis and interviews of traditional doctors (the holders of African medicine knowledge) were performed to gather sufficient data to achieve the work. Findings – The paper addresses this question of how to build a practical large-scope computer-aided diagnosis and prescription system which can exploit deep descriptions of ATM concepts, including mystical considerations. The system also provides scientific interpretations to some concepts sometimes considered as mystical facts. It is a java web-based platform combined to a Java Agent Development framework multi-agent system accessing an ontology to provide its results. Research limitations/implications – Because of the origins of healers involved in this research (from Gabon and Cameroon, countries of Central Africa), the ontology and the collected data may lack generalizability in the African scope and then it is a prototype. Therefore, ATM experts all over the continent are encouraged to participate to improve and standardize the ATM ontology and to populate the knowledge base. On the other side, the system cannot give scientific explanations to all the mystical considerations in ATM, there still some facts which cannot be rationally explained for now. Practical implications – The paper demonstrates the practical usability of the implemented system on the diagnosis and the treatment of a patient case. Social implications – The research describes a system which once validated by traditional experts, will serve as a tool to assist them in their day-to-day diagnosis and prescription tasks and will also serve as a reference on ATM practices for all interested users. Originality/value – The paper provides an in-depth description of a computer-aided diagnosis system (CADS) that promotes indigenous technology from an African perspective. Comparing to the former systems identified in the literature, the proposed system is the first which deals with believes and mystical considerations in ATM, and also the first which provides a function to rank its results.

Download Full-text

Integrating process modelling into multi-agent system engineering

Multiagent and Grid Systems ◽

10.3233/mgs-2012-0182 ◽

2012 ◽

Vol 8 (1) ◽

pp. 105-124 ◽

Cited By ~ 14

Author(s):

Tobias Küster ◽

Marco Lützenberger ◽

Axel Heßler ◽

Benjamin Hirsch

Keyword(s):

System Engineering ◽

Process Modelling ◽

Multi Agent System ◽

Agent System ◽

Multi Agent ◽

Integrating Process

Download Full-text