Projects for Research Students at Monash
1. Artificial Intelligence Applications
2. Intelligent Multicriteria Decision Support Systems
To develop intelligent multicriteria decision making models that
can be implemented as a decision support system for solving the general
multicriteria decision problem.
Specific intelligent procedures
to be developed for improving the decision quality include
* Context-dependent validation
techniques for decision outcomes.
The validation of the models used
will be examined in terms of the degree to which they are consistent with the
preference structures and decision information embedded in the decision
problem.
* Consensus under multi-panel
decision environments.
A consensus formation procedure
with a what-to-do analysis will be developed to explore the multiple
directions, among which satisfactory compromises can be reached. Some form of
consensus will be reached as to which of these are strongly desirable, or
strongly undesirable, in each context.
* Determination of the best
selection policy.
This procedure generates the
optimal policy for selecting alternatives on an on-going basis in a given
context. Dynamic programming and fuzzy clustering techniques will be considered
and compared with multicriteria analysis techniques.
Fuzzy knowledge bases for
handling the uncertainty of the decision process in different contexts will be
developed, including:
* Fuzzy knowledge bases for criteria weighting with subjective and
uncertain preferences
- Link the process of determining criteria weights with the
objectives of the decision problems, resulting in a consistent approach to
criteria weighting.
* Fuzzy knowledge bases for the selection and ranking of decision
alternatives
- Provide intelligent decision analysis and advice to the decision-makers
according to their preference structures and cognitive style, resulting in
informed decision outcomes with effective sensitivity analysis capabilities.
* Fuzzy knowledge bases for the validation of decision
outcomes
- Support the model selection procedure that best reflects the
decision information content, and provide experience in the implementation.
The project is intended for PhD research, although part of it may
be taken initially as a Masters/Honours research.
3. Composite
Index Construction with Optimal Weighting of Interactive Indicators
To
develop multicriteria decision making (MCDM) methods for constructing composite
indices from interactive indicators in an optimal way.
MCDM has
been regarded as a primary technique for constructing composite indices, which
are widely used for performance measurement based on multiple sources of
indicators. This research aims to address some challenging issues in MCDM for
constructing composite indices, including
*
Indicator interactions:
Fuzzy
cognitive maps will be constructed to model indicator interactions with a
network structure, while fuzzy measures with the Shapley value will be used to
model indicator interactions with a hierarchical structure.
*
Indicator weighting:
Optimal
weighting models will be developed to address the question of how to best
weight interactive indicators.
*
Aggregation model validation:
Context-dependent
validation techniques will be developed to address the question of how to
validate and select multicriteria aggregation models when the right answer is
unknown.
New
methodologies developed will be applied to performance comparison and ranking
problems involving quantitative or objective data such as city sustainability
and city mobility, and to alternative evaluation and selection problems
involving qualitative data and subjective assessments such as airline corporate
social responsibility and project innovation.
4. Fuzzy Multicriteria Classification
To develop algorithms for classifying items based on multiple
criteria with fuzzy or imprecise information.
The ABC inventory analysis or student performance analysis can be
used as a case study.
5. Risk Analysis and Evaluation under Uncertainty
To develop risk management techniques and models for
airline/airport safety or other business operations.
Major research tasks
(methodology) include:
* Identifying and classifying risk factors (Fuzzy AHP survey
questionnaire and fuzzy clustering).
* Generating a risk index based on the degree of importance,
probability, and gravity of hazard (Risk grid).
* Determining the threshold values of risk and degree of
tolerable risk severity (As low as reasonably practicable principle).
* Evaluating treatment alternatives (Fuzzy MCDM).