Information Technology Reference
In-Depth Information
We acknowledge that this description may not provide the reader with a complete
understanding of our ABM. Space restrictions preclude providing greater detail. In lieu,
the ABM is available as a zip file. The interested reader can download it from
http://www.uai.cl/images/stories/CentrosInvetigacion/CINCO/CAT_1_English.zip
. To run it, you
will need first to install Netlogo version 4.0.4 (http://ccl.northwestern.edu/netlogo/). Once
on the ABM interface, do the following steps to run a simulation: (1) Input the simulation
parameters, be it with the slider controls or typing the desired
p(a1)
and
p(a2)
values in the
appropriate windows. (2) Press SETUP. (3) Press SIMULATE. (4) If you want to pause a
run, simply press SIMULATE (you will need to press it again to resume).
4
Preliminary Results
Once the ABM was implemented and verified, we carried out several runs to assess
the dynamics of the coefficients of the concepts that emerged. After gaining some
insights into the dynamics of those coefficients and how different combination of
parameters changed those dynamics, we performed experiments fixing the value of
some parameters as follows: number of agents = 100; number of versions of a concept
(or number of related concepts) = 5; initial value of coefficients of concepts = 0.5; and
changing the value of
p(a1)
and
p(a2)
between 0.1 and 0.95. According to the
combination of values for
p(a1)
and
p(a2)
, three different dynamics emerged.
4.1
Convergence to Zero
When we set small values for
p(a1)
and
p(a2)
, for example
p(a1)
= 0.1 and
p(a2)
=
0.2, then the coefficients rapidly decrease and get close to zero, remaining at that
value. This can be seen in Figure 1, where we plotted the mean value of the
coefficient of each version of the concept (c1 through c5) over simulation steps. The
mean value of each coefficient is calculated by averaging the individual value of
the coefficient of the agents that have each version of the concept.
Fig. 1.
Mean of coefficients of concepts (c1 through c5), for
p(a1)
= 0.1 and
p(a2)
= 0.2
