Table 20.3 Provider's constraints
10 - 11
11 - 12
13 - 14
14 - 15
a lower. Note that there is not a direct correlation between the value of an
attribute and its associated rank. In some cases it might occur that the
greater the value, the greater the rank but this is not always the case.
Each attribute is associated with a weight that expresses its relevance in
the negotiation process. For example, if the technician responsible for the
laboratory can accept a change in computer memory better than a change in
the time, then the time has a higher weight than the memory. Attributes
with a single admitted value do not have a corresponding weight (i.e. they
are not negotiable), because that value is required. Table 20.3 represents a
synthetic way of expressing preferences for attribute values. The alternative
would be to enumerate all the possible combinations of attributes and values
and to order them according to the user preferences.
A contract proposal consists of the selection of a single value for each
attribute. Let R i , (for i # 1, 2, 3, ...) be the rank associated to the value that
attribute i assumes in the contract proposal during negotiation step k (for
k # 1, 2, 3, ...), W i the weight of attribute i , and T k the total value associated
to a contract proposal at step k . The following relations hold:
W i =
1 T k =∑
W i +
Equation 20.1 Proposal evaluation
Using Equation 20.1, the proposer can rank every admissible contract and
select the best proposal to submit to the reviewer. More sophisticated
methods for enumerating and ranking admissible contracts can be found in
Triantaphyllou and Shu (2001).
This approach resembles a typical weighted sum multi-criteria model: the
entire contract proposal is reduced to a single number - the total value.
The basic negotiation mechanism is contract matching. At the beginning
of the negotiation process, the proposer and the reviewer exchange contract
proposals that have different values for some or all of the attributes.
During each negotiation step, the reviewer (in our case the laboratory
server) evaluates the proposer's contract and formulates a counter-proposal.
The simplest approach consists of selecting a contract that reduces the