Authors:
Ana Tapia-Rosero
1
and
Guy De Tré
2
Affiliations:
1
FIEC and Ghent University, Ecuador
;
2
Ghent University, Belgium
Keyword(s):
LSP, Evaluation, Relevant Opinions, Shape-similarity Method, Large-scale Group Decision-making.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Computational Intelligence
;
Fuzzy Information Processing, Fusion, Text Mining
;
Fuzzy Systems
;
Pattern Recognition: Fuzzy Clustering and Classifiers
;
Soft Computing
;
Soft Computing and Intelligent Agents
Abstract:
We propose to identify which opinions are relevant, from the decision-maker’s point of view, within a large group of opinions that could be collected using social media. Our approach considers that each participating person expresses his/her preferences over a criterion specification as a matter of degree. First, using a shape similarity method, we split a large group of opinions, where each opinion is represented through a membership function, into clusters —here, a cluster depicts a group of similar opinions over the criterion. Then, in order to evaluate the relevance of each cluster, we differentiate them based on some characteristics like the cohesion, the number of membership functions and the number of noticeable opinions. Within this paper, the cohesion of the cluster is a measure that takes into account the level of togetherness among its contained membership functions; and the representativeness of the cluster is obtained by combining the number of membership functions and t
he number of noticeable represented opinions (i.e., considered as more important or worthy of notice among other opinions). Moreover, relevant clusters result in the evaluation of combining their cohesion measure and their representativeness according to the decision-maker’s point of view. Finally, as a part of the evaluation, this proposal includes the steps describing the process through an illustrative example.
(More)