Lucas Marin

SigTur/E-Destination: Ontology-based personalized recommendation of Tourism and Leisure Activities

SigTur/E-Destination is a Web-based system that provides personalized recommendations of touristic activities in the region of Tarragona. The activities are properly classified and labeled according to a specific ontology, which guides the reasoning process. The recommender takes into account many different kinds of data: demographic information, travel motivations, the actions of the user on the system, the ratings provided by the user, the opinions of users with similar demographic characteristics or similar tastes, etc. The system has been fully designed and implemented in the Science and Technology Park of Tourism and Leisure. The paper presents a numerical evaluation of the correlation between the recommendations and the users motivations, and a qualitative evaluation performed by end users.

On-line dynamic adaptation of fuzzy preferences

Recommender systems are very useful in domains in which a large amount of continuous information needs to be evaluated before a decision is made. Systems that permanently interact with users need to be adapted to changes in their interests. This paper proposes an algorithm that takes advantage of the preference information implicit in the actions of the user to dynamically adapt the user profile, in which user preferences are represented as fuzzy sets. The algorithm has been tested with real data extracted from the New York Times and has shown promising results. This paper presents the adaptation algorithm and discusses the influence of its basic parameters.

Automatic preference learning on numeric and multi-valued categorical attributes

Learning preferences implicitly is a challenging task in the design of recommenders.Our approach infers preferences by analyzing choices without any explicit feedback.Choices considered are defined by numerical and multi-valued categorical criteria. One of the most challenging tasks in the development of recommender systems is the design of techniques that can infer the preferences of users through the observation of their actions. Those preferences are essential to obtain a satisfactory accuracy in the recommendations. Preference learning is especially difficult when attributes of different kinds (numeric or linguistic) intervene in the problem, and even more when they take multiple possible values. This paper presents an approach to learn user preferences over numeric and multi-valued linguistic attributes through the analysis of the user selections. The learning algorithm has been tested with real data on restaurants, showing a very good performance.

Knowledge-driven delivery of home care services

Home Care (HC) assistance is emerging as an effective and efficient alternative to institutionalized care, especially for the case of senior patients that present multiple co-morbidities and require life long treatments under continuous supervision. The care of such patients requires the definition of specially tailored treatments and their delivery involves the coordination of a team of professionals from different institutions, requiring the management of many kinds of knowledge (medical, organizational, social and procedural). The K4Care project aims to assist the HC of elderly patients by proposing a standard HC model and implementing it in a knowledge-driven e-health platform aimed to support the provision of HC services. This paper focuses on two knowledge-based personalization aspects incorporated in the platform that aim to overcome the difficulties of HC delivery. The first one is the assistance to medical practitioners in the process of defining a customized treatment adjusted to the medical and social conditions of a particular patient in order to consider multiple co-morbidities. The second one is the possibility of tailoring the profiles of the care professionals according to the medical and organizational daily requirements in order to allow a flexible care delivery. Those two aspects, guided by the knowledge explicitly represented in the platform, play a crucial role in the medical and social acceptance of this kind of e-health systems. The paper also includes a real case study designed and tested by healthcare professionals and includes encouraging results from the test of the platform in a real health care environment in the city of Pollenza (Italy).

Dynamic adaptation of numerical attributes in a user profile

Recommender systems try to help users in their decisions by analyzing and ranking the available alternatives according to their preferences and interests, modeled in user profiles. The discovery and dynamic update of the users’ preferences are key issues in the development of these systems. In this work we propose to use the information provided by a user during his/her interaction with a recommender system to infer his/her preferences over the criteria used to define the decision alternatives. More specifically, this paper pays special attention on how to learn the user’s preferred value in the case of numerical attributes. A methodology to adapt the user profile in a dynamic and automatic way is presented. The adaptations in the profile are performed after each interaction of the user with the system and/or after the system has gathered enough information from several user selections. We have developed a framework for the automatic evaluation of the performance of the adaptation algorithm that permits to analyze the influence of different parameters. The obtained results show that the adaptation algorithm is able to learn a very accurate model of the user preferences after a certain amount of interactions with him/her, even if the preferences change dynamically over time.

The Unbalanced Linguistic Ordered Weighted Averaging operator

Aggregation operators for linguistic variables usually assume a uniform and symmetrical distribution of the linguistic terms that define the variable. A well-known aggregation operator is the Linguistic Ordered Weighted Average (LOWA), which has been extensively applied. However, there are some problems where an unbalanced set of linguistic terms is more appropriate to describe the objects. In this paper we define the Unbalanced Linguistic Ordered Weighted Average (ULOWA) on the basis of the LOWA operator. ULOWA takes into account the fuzzy membership functions of the terms during the aggregation process. There is no restriction on the form of the membership functions of the terms, which can be triangular or trapezoidal, non symmetrical and non equally distributed. The paper demonstrates the properties of ULOWA. Finally, a comparison of this operator with some other aggregation operators for unbalanced sets of terms is done.

Personalised recommendations based on novel semantic similarity and clustering procedures

Intelligent data analysis methods usually require as input a matrix, in which each row is an object to be analysed and each column is an attribute. In most cases it is assumed that attributes are Boolean, categorical or numerical. With the advent of semantic domain information in the form of ontologies, it is now common to find also semantic attributes, which may take as value a list of concepts. This paper proposes a new ontology-based procedure to compute the similarity between lists of semantic values, which may be used to compare objects. This measure is employed in an enhanced version of the k-means clustering method. The usefulness of the obtained classes has been tested in the context of a Web-based personalised recommender of Tourist destinations.

Induced Unbalanced Linguistic Ordered Weighted Average and Its Application in Multiperson Decision Making

Linguistic variables are very useful to evaluate alternatives in decision making problems because they provide a vocabulary in natural language rather than numbers. Some aggregation operators for linguistic variables force the use of a symmetric and uniformly distributed set of terms. The need to relax these conditions has recently been posited. This paper presents the induced unbalanced linguistic ordered weighted average (IULOWA) operator. This operator can deal with a set of unbalanced linguistic terms that are represented using fuzzy sets. We propose a new order-inducing criterion based on the specificity and fuzziness of the linguistic terms. Different relevancies are given to the fuzzy values according to their uncertainty degree. To illustrate the behaviour of the precision-based IULOWA operator, we present an environmental assessment case study in which a multiperson multicriteria decision making model is applied.

Induced Unbalanced Linguistic Ordered Weighted Average

Aggregation operators for linguistic variables usually assume a uniform and symmetrical distribution of the linguistic terms that define the variable. This paper de- fines the Induced Unbalanced Linguistic Ordered Weighted Average (IULOWA). This aggregator takes into account the fuzzy membership functions of the terms during the aggregation operations of the pairs of terms. There is no restriction on the form of the mem- bership functions of the terms, which can be triangular or trapezoidal, non-symmetrical and non-equally dis- tributed. Moreover, the paper proposes to use the speci- ficity and fuzziness measures of the terms to induce the order of the arguments, providing some examples of this criterion in decision making.

Dynamic Learning of Keyword-Based Preferences for News Recommendation

The accurate recommendation of daily news requires a detailed knowledge of the topics of interest to the user. The dynamic and continuous analysis of the content of the news that are read (or ignored) by the user every day may lead to the automatic, unsupervised and non-intrusive learning of the positive (and negative) preferences of the user with respect to a set of keywords. These preferences may then be used to rank the daily news, so that the user is recommended those items that match better with his/her interests. The cyclic preference learning methodology described in this paper is illustrated with a case example based on real news from the British newspaper The Guardian, in which promising results have been obtained.