2 PROPUESTA PARA LA PLANIFICACIÓN DE LA DEMANDA
3.1 Operaciones Previas
3.1.1 Especificaciones de los requerimientos
63 3.1.7 Summary
Figure 15 – STH overview
Figure 15 illustrates a summary of STH methodology, it is visible that the Blue Hat is not in the centre of the figure, although it plays a crucial role in the meeting, because any participant besides the moderator can intervene in a blue-hat way.
The method of the six hats to think was designed to remove thought from habitual argumentative style and to take it to a cartographic style. This is done by process of two stages: first it is to develop the map; second it is to choose the route. If the map is sufficiently good, a better route usually is obvious.
The greater value of the hats is its own artificiality. It offers a formality and one convention to, as much as possible, require a certain type of thought from every participant. STH lays down the rules of the game of the thought. Whoever plays it is going to know these rules – and people are generally good at following rules and playing games.
3.2 ALBidS Methodology based on STH
This work’s focus is creating a decision support tool that acts based on STH method. For this development, it was decided to use ALBidS architecture while adapting agent strategies to each different “hat thinking way”. This means that the scope of this support will be solely for bidding in the Spot Market.
It was also necessary to relate MASCEM and ALBidS’ internal working to that of a common STH method-driven meeting, mainly these characteristics:
A moderator entity, leading different interveners in the decision – the blue hat or the moderator sets the agenda for the meeting, determines its closing and final decision.
64
Different ways of thinking about the problem to solve
No arguing or debating, no confrontation
Figure 16 depicts the proposed architecture for this system. The approach that was taken was to build STH as an ALBidS’ strategy agent. Enacting the STH’s Blue Hat role as moderator, this agent will be responsible for controlling the process and ruling the final decision that will be communicated to ALBidS.
Figure 16 – STH Architecture
As denoted before, the STH method proposes to end or mitigate any conflict or confrontation that is natural to occur in a meeting (at least in Western Civilizations [De Bono E., 1985]);
however unlike a human meeting, in this system it won’t be necessary to avoid conflict or arguing because there is no communication or negotiation between.
Unlike STH where each intervenient is asked to wear every hat, in our approach, each intervenient will only act out a role and this will be played by an ALBidS’ strategy agent. For this purpose we mapped a relation between each Hat’s behaviour and the existing ALBidS’
strategies, agents or mechanisms.
Also, in order to take advantage of what was already done, we chose not to try to assemble a exactly STH-like “thinking map” to where all interveners would contribute with their part (feasible with a blackboard approach, for example); [De Bono E., 1985] states that after our
“thinking map” is done, the solution will be pretty obvious to every participant in the meeting.
Because we are dealing with multiagent systems and concepts such as “pretty obvious” and
“emotions” are unknown (or hard to develop) to these artificial entities, we had to make
3.2 ALBidS Methodology based on STH
65 some compromises and decisions on what would represent these concepts. The reasons that led to the Hat-Agent relation were:
White Agent – this hat is responsible for bringing data, figures, and numbers into discussion, as such little or no treatment should be done in the data; it was clear that we could use an Average approach for this role.
Red Agent – we decided to examine the emotions as something related to the recent past. Though not a true emotional decision or fulfilling the role De Bono planned for this hat, it is reasonable to assume a human intervenient could say something like
“Due to our recent past, I’ve got a hunch that we should not raise prices”; a regression approach was chosen for this role.
Black Agent – this role is for caution, to avoid danger. It is quite fair to think of it as knowing our position in the market, and knowing what we can or can’t do; this is why we chose to fill this role with an Economic Analysis with little or no propensity for risk.
Yellow Agent – Optimism without foolishness, this is the best way to describe the yellow hat thinker; also as the opposite of the black view, we saw this agent as a logical analyser with some appetite for risk – hence, once again the Economic Analysis this time with medium to high risk.
Green Agent – the evolutionary traits of Particle Swarm Optimization used in Determinism Theory Agent made this strategy a suitable candidate for the creative thinking part. We are looking for logical and adjusted decisions while somehow different from the usual.
Blue Agent – the one responsible for the development of the process and the final decision. This is the STH agent, the one that will gather every other agent decisions, summarize, and deliver a final answer to ALBidS. This agent’s capabilities and structure is fully analysed in the section 3.3.
Table 1 shows the final version of this relation.
Table 1 – Relation between STH’s roles and existing ALBidS’ entities STH’s Role ALBidS’ Existing Agent
White Average 1 Agent Red Regression 2 Agent
Black Economic Analysis Agent – with low risk
Yellow Economic Analysis Agent – with medium-high risk Green Determinism Theory Agent – using PSO
Blue Not Available
Each of these Strategy Agents will be, henceforth and on the context of STH, referred to by the colour of its respective hat – in other words Red Hat Agent will refer to Average 2 Agent – or collectively as STH Agents.
66
Another compromise, that was already referred, was the rejection of using a collaborative blackboard system to build our “thinking map”. Though this would be the natural approach, it would have forced us to develop an independent MAS, and with it the development of another facilitator beyond those already present in MASCEM (OAA) and the Prolog Facilitator of ALBidS.