19.11.2013 Views

Untitled - Universidade de Lisboa

Untitled - Universidade de Lisboa

Untitled - Universidade de Lisboa

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Innovation Diffusion in Organizations- An Evolutionary Approach<br />

Luciano Sampaio 1 , João Varajão 2,3 , E.J. Solteiro Pires 2,4 , P.B. <strong>de</strong> Moura Oliveira 2,5<br />

1<br />

Instituto <strong>de</strong> Estudos Superiores <strong>de</strong> Fafe, Lda. - Escola Superior <strong>de</strong> Tecnologias <strong>de</strong> Fafe, Portugal<br />

2 University of Trás-os-Montes e Alto Douro, Vila Real, Portugal<br />

3 Centro ALGORITMI, Guimarães, Portugal<br />

4 CITAB Research Center, Vila Real, Portugal<br />

5 CIDESD Research Center, Vila Real, Portugal<br />

luciano.sampaio@optimus.clix.pt, jvarajao@utad.pt, epires@utad.pt, oliveira@utad.pt<br />

Exten<strong>de</strong>d abstract<br />

Over the last years several theoretical mo<strong>de</strong>ls have been proposed to explain the innovation adoption<br />

process, each one using different approaches [1-2]. Therefore, various mathematical mo<strong>de</strong>ls, based on<br />

simple expressions have been <strong>de</strong>veloped to <strong>de</strong>scribe the diffusion of an innovation among a given<br />

potential adopters set. Today, these mo<strong>de</strong>ls are a fundamental tool to predict the innovation adoption<br />

growth. Several examples can be found in literature, as the Gompertz and Bass macro-logistics mo<strong>de</strong>ls<br />

[3], among many others.<br />

The application of these mo<strong>de</strong>ls is ma<strong>de</strong> accordingly to the estimation of certain coefficients. These<br />

coefficients are typically estimated from time series. However, in several problems there is not any<br />

historical data, which constitutes a serious drawback in applying this type of methods. Thus, to<br />

simulate the process sometimes is necessary to use features of other similar innovations [3, 4]. This<br />

could introduce some entropy in the process. Moreover, the mo<strong>de</strong>ls based on the logistic equation have<br />

a limited application. On the other hand, mo<strong>de</strong>ling at macro level is inaccurate because it assumes a<br />

perfect social mix where everyone interacts with everyone and does not measure whether the<br />

interpersonal are the same [4]. Other type of mo<strong>de</strong>ls has been proposed, based on the study of the<br />

individual and / or organization within a social system [5-6]. Some approaches consi<strong>de</strong>r that the<br />

diffusion is ma<strong>de</strong> through the influence of social network. In this case, the influence is transmitted by<br />

direct contact between individuals [4,7]. For instance, Young [7] proposed and built a mo<strong>de</strong>l based on<br />

a graph.<br />

Taking the former work in consi<strong>de</strong>ration, particularly the Young mo<strong>de</strong>l [7], and evolutionary<br />

computation [8], this work proposes a new mo<strong>de</strong>l using evolutionary algorithms to <strong>de</strong>scribe the<br />

innovation adoption process. Moreover, the proposed mo<strong>de</strong>l un<strong>de</strong>r <strong>de</strong>velopment is based in Roger,<br />

Valente and Davis mo<strong>de</strong>ls concepts [5,4,9] and Young interaction network [7]. The main advantage of<br />

a mo<strong>de</strong>l with these features is that historical data is not required to simulate the innovation diffusion,<br />

since its construction is focused on the individual characteristics, in their behavior towards innovation,<br />

external influences (media) and their interpersonal relationships. Moreover, it only requires taking into<br />

account the individual adoption trends and social systems (businesses, schools, municipalities, etc.)<br />

converted into mathematical mo<strong>de</strong>ls, as well as the network of direct contacts between individuals. The<br />

predisposition of an individual to adopt a particular innovation can also be evaluated probabilistically.<br />

5

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!