Construction of an Outranking Relation Based on Semantic Criteria with ELECTRE-III
ELECTRE-III is a well-known multi-criteria decision analysis method that ranks a set of alternatives in terms of a set of heterogeneous evaluation criteria. It is based on constructing and exploiting a pairwise outranking relation between alternatives, wh
- PDF / 842,016 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 15 Downloads / 167 Views
ract. ELECTRE-III is a well-known multi-criteria decision analysis method that ranks a set of alternatives in terms of a set of heterogeneous evaluation criteria. It is based on constructing and exploiting a pairwise outranking relation between alternatives, which are defined with numerical and ordinal values. However, nowadays it is very common the use of descriptive linguistic tags, which is information that requires a qualitative treatment rather than a numerical one. In this paper we propose to store the user preferences about a set of tags in an ontological structure and to use this knowledge to construct the outranking relation by means of a semantic analysis of the tags associated to the alternatives. Uncertainty is handled by means of fuzzy concordance and discordance functions. The method is illustrated with a case study related to the recommendation of touristic activities. Keywords: MCDA
Ontology Semantic profile Outranking relations
1 Introduction The problem of ranking a set of alternatives has been thoroughly studied in MultiCriteria Decision Aid (MCDA). Some well-known ranking methods, like the ELECTRE (ELimination and Choice Expressing REality) family [2], construct a preference structure from a pairwise comparison of a set of possible alternatives, which is based on two voting-inspired ideas: concordance (or “the choice of the majority”) and discordance (or “the respect to minorities”). These outranking techniques have been widely used in many fields [4]. An important advantage of these methods is that they can work directly with purely ordinal scales, without requiring their transformation into abstract ones with an arbitrary range. A second advantage is that indifference and preference thresholds can be used to model uncertain knowledge. However, one of the main shortcomings of ELECTRE is that alternatives can only be defined in terms of numerical and ordinal criteria. Nowadays it is becoming increasingly common to find decisional situations in which alternatives may also include non-numerical information, represented in the form of semantic criteria, which may take as values the concepts of a given domain ontology. For example, the description of a touristic destination may include numerical © Springer International Publishing Switzerland 2016 J.P. Carvalho et al. (Eds.): IPMU 2016, Part II, CCIS 611, pp. 238–249, 2016. DOI: 10.1007/978-3-319-40581-0_20
Construction of an Outranking Relation Based on Semantic Criteria
239
criteria (e.g. number of inhabitants, average temperature) but also semantic ones (e.g. sports that may be practiced in the city, cultural attractions that may be visited). Ontologies are knowledge structures that commonly store the main concepts of a domain, the taxonomic and non-taxonomic relationships between them, and their attributes [8]. A semantic user profile [10] may contain the degree of preference of the user with respect to some domain concepts. This information may be exploited to compare and rank a set of alternatives. The main aim of this paper is to show how
Data Loading...
