|
1. Research
|
See Research Theme and Topics
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
|
Research Theme and Topics
|
|
|
I have established the ECE Department's Intelligent Software Systems (ISS) Laboratory. The following map depicts categories of research and development that ISS is engaged with. My specialization is in engineering of intelligent, distributed and heterogeneous networked systems, specifically in design and implementation of agent-based software systems and support tools and techniques for groupware systems. These activities have been categorized depending largely on software and agents. My research activities cover a variety of topics including the followings:
- Uncertainty and hostility management in multi-agent systems:
The goal of this research is to address competition within an organization, in which knowledge sharing is impossible. We develop and present an incomplete game theoretical based decision making method for competitive agents. (Received 2 best paper awards) (Funded by NSERC).
- Agent-based Software Engineering (Agent-SE) Methodology:
In this research we formalize the development process of multiagent systems as well as the knowledge representation and sharing of agents for cooperative and coordinative agents and use the results in large scale multi-agent system design. (2 keynote lectures)
- Methodological support for interactive software agents:
The goal of this research is to devise theories, techniques and measures for enhancing quality and reliability of agent-based systems. Unique points with this research are (a) focus on agent system reliability and quality; (b) focus on agent interactions by starting with a complete set of possible interactions among software agents; and (c) focus on decision making based on multiple threads of control rather than reasoning based on a single thread of control.
- Distributed Software Agents for Network Fault Management:
This research offers a radically different local solution to network fault management as opposed to the centralized techniques that are commonly used in practice. The agents on local nodes are responsible for collecting data, correlating them using a Bayesian network and report only major events to the control station. (Received 2 best paper awards) (Funded by IPA, Japan).
- Distributed Multiagent Learning and Tutoring System based on Learning Ecology:
The goal of the research is to develop an intelligent tutoring system (ITS) that adapts the delivery of instruction according to the learner’s needs, by taking into account learner’s motivation states. Due to computational convenience, many other systems rely only on the learner response to exercises to assess his/her needs. In our approach, however, we looked at one step deeper, the learner’s learning drives, in order to find out what parameters affect the willingness to engage in learning.
- Enterprise Knowledge Management Using Knowledge Orchestration Agency:
In this research, knowledge orchestration is equivalent to peer-to-peer knowledge management using models of organizational interactions characterized by three concepts: flow of information among individuals (people or departments), decentralized management and modularization. The agents work on a body of information and have reasoning and representation mechanisms appropriate for that type of information. Reusable modules (library, package, subsystems, COTS, etc.) for representing and reasoning with the body of information will further reduce the long term operational costs.
- Intelligent project lifecycle knowledge management and decision support:
Mission critical decision making in enterprises depends heavily on intelligent systems for extracting, analyzing and interpreting information from multiple heterogeneous, distributed data and knowledge sources. It is assumed that data warehouses (DW), data marts (DM) are required for optimized data accessibility and use. This research propose a novel architecture based on multiagent technology to support information and knowledge extraction over distributed data sources in order to use them in the decision making process. The proposed framework is applied to a real-world project lifecycle case that is EPC (Engineering Procurement and Construction) project.
- Commercial off-the-shelf software components evaluation method using multiagent technology:
The target problem for this research, i.e., COTS selection, is selected to bring together the agent-based and decision support methodologies. This provides us with a test bed for methods of interaction (e.g., negotiation, competition, etc.) and traditional decision theories (e.g., maximum utility, game theory, etc.). The target has a very high potential for commercialization.
- Knowledge management in automatic software design:
In Software Creation project we build a computer aided software engineering (CASE) tool that can imitate the design steps of human designers. There are two types of knowledge involved in human design: detailing knowledge represented by conversion and detailing rules; and the knowledge required for hierarchical expansion, represented by micro design rules. The former refers to the design product knowledge and the latter maps to the design process knowledge. The system incorporates both types of knowledge. (Received 1 best paper award)
- Methodologies for automating transition from stakeholders' requests to skeleton code:
In this project we focus on use-case driven object elicitation, class modeling and skeleton code generation based on natural language requirements and tailor it to the IBM’s Rational Extended Development Environment (XDE). We implement a couple of tools and plug-ins that enhance the functionality of XDE by accepting natural language requirements as their input and generating skeleton code as their output. The whole process is divided into two parts based on two main concerns. The first part, called Use Case Model Generator (UCMG), addresses natural language (NL) requirements analysis and use-case modeling. The second part, called Use Case driven Development Assistant (UCDA), is concerned with the use-case realization, object/class elicitation and skeleton code generation.
- Intelligent Software Measurement System:
Software measurement, in order to be effective, must be focused on specific goals; applied to all life-cycle products, processes and resources; and interpreted based on characterization and understanding of the organizational context, environment and goals. The Goal-Question-Metric (GQM) was developed in response to the need for a goal-oriented approach that would support the software measurement. In this research the Intelligent Software Measurement System (ISMS) is developed following the goal-driven software measurement process. In the ISMS project we automate the 10 steps of the process. The main development tasks of ISMS are eliciting the knowledge and experience from software measurement experts, representing it in a flexible yet well-structured way, and building a knowledge base infrastructure for the system.
Old research topics:
- Intelligent Agents for Electronic Commerce
- Qualitative Functional Reasoning
- Qualitative Sensitivity Analysis
- Qualitative Reasoning in Supervisory Control
Please click on Research buttons on the left to go to research page. Or click on a link below to get a full description of each theme. Note that some of the files may have restricted access and may be for local use only.
|
|
|
