TAMRI is a planning tool for identifying task assignments based on multiple criteria and weighted project goals.Its implementation combines a distributed systems approach with Bayesian networks. The tool can be adapted to specific organizational environments by exchanging the underlying Bayesian network.The tool implements an algorithm that uses (1) a set of weighted goals, (2) characteristics of the project tasks and the available sites, and (3) causal relationships between characteristics and goals for deriving a weighted list of suggestions for task allocations.
Implementation algorithm:-
For every task and every site:
- Instantiate a BN for execution costs based on project and resources characteristics
For every combination of two tasks and two sites:
- Instantiate a BN for communication costs
For n times (n >> 1, e.g., n = 2000):
- Select random values for every goal in every BN according to the probability distributions
- Integrate the different goal values into one cost value per BN (normalizing, weighting, and adding)
- Execute the distributed system algorithm with the execution cost and transmission cost values from the previous step
- Add the returned assignment to the list of optimal assignments
Return the list of assignments in decreasing order of occurrences
Although the toll seems to very useful but there were few limitation. The people using the tool must be an expert and have very good knowledge in distributed project development. Determining and quantifying the nature of resources was another big issue.Lesser the knowledge about the project more uncertain the results are.
No comments:
Post a Comment