Observatorio de I+D+i UPM

Memorias de investigación
RIDER- A recommendation framework for exploiting evaluation results and user quality requirements
Research Areas
  • Information technology and adata processing
Analytic Network Process and Analytic Hierarchy Process are Multiple Criteria Decision Making methods which are often used for providing recommendations in terms of rankings in order to help with the problem of selecting one product, service or action among a group of possible alternatives. The usage of the Analytic Network Process and of the Analytic Hierarchy Process implies a definition of a model of a recommendation, and involvement of persons (users or experts in the field related to the alternatives to be recommended) for the manual comparison of alternatives according to their quality by specifying preferences (importance) of one alternative over another. However, different users have different quality requirements over alternatives and every specific set of quality requirements implies repeated definition of a model of a recommendation adapted to that set of quality requirements, which is a resource consuming task. Furthermore, relying on user or expert comparison of alternatives is often expensive since it requires knowledge about all the alternatives and about the comparison method used with the Analytic Hierarchy or Network Process. On the other hand, in some fields in which the selection of the best alternative is a difficult task, such as the field of Semantic Web and semantic technologies, there exist a number of alternatives? quality evaluation results that can be used for the comparison of alternatives and deriving preferences between these alternatives. Such evaluation results are usually obtained in the process of evaluation driven by the guidelines defined in a quality model. This thesis aims to advance the current state of the art in the Multiple Criteria Decision Making field and in the semantic technologies field. In particular, the main goals of this thesis are: i) to provide a Multiple Criteria Decision Making framework that extends the Analytic Network Process and that takes advantage of evaluation results of alternatives and of user quality requirements over such alternatives; and ii) to apply this MCDM framework in the Semantic Web field for semantic tools recommendation. To achieve these goals, the following contributions are delivered in this thesis: _ A set of domain-independent algorithms for the automatic comparison of alternatives according to alternatives? evaluation results, which can be used in the Analytic Network Process and in the Analytic Hierarchy Process. _ A set of methods for the dynamic extraction of the Analytic Network Process and Analytic Hierarchy Process models, that are based on user quality requirements. _ Software that supports the proposed MCDM framework. _ SemQuaRE, a quality model for semantic tools. _ The application of the proposed MCDM framework for the recommendation of semantic technologies, including a web application for the recommendation of such technologies. The MCDM framework proposed in this thesis is generic and can be instantiated in any domain by defining a quality model, an AHP or ANP model of the domain, criteria pairwise comparisons (as required by the AHP or the ANP) made by experts, and taking evaluation results of alternatives as inputs. Once instantiated, the proposed framework takes a set of quality requirements specified by a user as input and produces as output a ranked list of alternatives that best suit the specified user quality requirements.
Mark Rating
Sobresaliente cum laude
  • Director: Raul Garcia Castro (UPM)
  • Director: Asuncion de Maria Gomez Perez (UPM)
  • Autor: Filip Radulovic . (UPM)
Research Group, Departaments and Institutes related
  • Creador: Grupo de Investigación: Ontology Engineering Group
  • Departamento: Inteligencia Artificial
S2i 2020 Observatorio de investigación @ UPM con la colaboración del Consejo Social UPM
Cofinanciación del MINECO en el marco del Programa INNCIDE 2011 (OTR-2011-0236)
Cofinanciación del MINECO en el marco del Programa INNPACTO (IPT-020000-2010-22)