Modeling Semantic Web Services by Learning from Users’ Feedback Francesca A. Lisi and Floriana Esposito Department of Computer Science Lab of Knowledge Acquisition and Machine Learning (LACAM) { francesca.lisi,floriana.esposito } @uniba.it ILP 2014 - Sept. 14, 2014 F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 1 / 15
Puglia@Service Generalities Fund: PON Research & Competitivity 2007-2013 Partnership: High-Tech District + Academia + Industry Area of intervention: Apulia Region, Italy Objective Internet-based service infrastructure for the Apulia Region, Italy Focus on Knowledge Intensive Services (KIS) Application areas 1 Public Administration 2 Integrated Tourism F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 2 / 15
Semantic Web Services Web Services + Semantic Web Server end of a client-server system for machine-to-machine interaction via the WWW Defined with mark-up languages which makes data machine-readable OWL-S Upper-level ontology for Semantic Web Services Compatible with the Ontology Web Language (OWL) Automated service discovery, invocation, composition and monitoring F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 3 / 15
Integrated Tourism Definition Kind of tourism which is explicitly linked to the localities in which it takes place and, in practical terms, has clear connections with local resources, activities, products, production and service industries, and a participatory local community. Aims 1 For the various interests, requirements and needs the aim is to be fused together into a composite, integrated strategic tourism plan. 2 For tourism the aim is to be planned with the intention of being fused into the social and economic life of a region and its communities. F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 4 / 15
An ontology for Integrated Tourism: OnTourism Metrics No. logical axioms: 196 No. classes: 115 No. object properties: 9 No. data properties: 14 Main entities Class Site (with subclasses, e.g. , Accommodation and Attraction ) Class Place Class Distance (with subclasses, e.g. , Distance by car and Distance on foot ) Properties hasLengthValue / hasTimeValue F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 5 / 15
Semantic Web Services for Integrated Tourism I Service1: destination attractions service i.p. Destination (from travel ) o.p. Attraction (from OnTourism ) Service2: city churches service i.p. City (from travel ) o.p. Church (from OnTourism ) Service2 specializes Service1 City subclass of Destination Church subclass of Attraction F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 6 / 15
Semantic Web Services for Integrated Tourism II Service3: near attraction accommodations service i.p. Attraction (from OnTourism ) o.p. Accommodation (from OnTourism ) Service4: wheelchairaccess accommodations service i.p. Wheelchair Access (from OnTourism ) o.p. Accommodation (from OnTourism ) Service5 as sequence of Service2, Service3, and Service4 It does return a list of wheelchair-accessible accommodations near churches in a given city, but .. It does not provide recommendations according to the user profile F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 7 / 15
Learning from Users’ Feedback The problem Learn inclusion axioms of the form C 1 and . . . and C n subclass of Bad Accommodation from positive and negative examples for Bad Accommodation (provided by users) and prior domain knowledge (ontology OnTourism ) The solution Foil - DL is a Foil -like method for learning under incompleteness and vagueness (Lisi & Straccia, ILP 2013) F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 8 / 15
Case study: Religious tourism in Bari I Festival of St. Nicholas F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ Feedback ILP 2014 - Sept. 14, 2014 9 / 15
Case study: Religious tourism in Bari II Pilgrim requirements Preference for accommodations near places of worship Need for wheelchair-accessible accommodations City requirements Preference for eco-mobility Sustainable touristic offer F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 10 / 15
Case study: Religious tourism in Bari III Dataset Population of OnTourism with data automatically extracted from TripAdvisor and GoogleMaps 34 hotels and 70 B&Bs 2 sites of interest (St. Nicholas’ Basilica and St. Sabinus’ Cathedral) 208 foot distances between accommodations and sites of interest with values in time and length F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 11 / 15
Case study: Religious tourism in Bari IV Setting 15 positive examples (% of positive users’ feedback below 0 . 5) 57 negative examples (% of positive users’ feedback over 0 . 7) Experiment 1: without distances from sites of interest Bed_and_Breakfast and hasAmenity some (Pets_Allowed) and hasAmenity some (Wheelchair_Access) subclass of Bad_Accommodation Experiment 2: with distances from sites of interest hasAmenity some (Bar) and hasAmenity some (Wheelchair_Access) and hasDistance some (isDistanceFor some (Bed_and_Breakfast) and isDistanceFor some (Church)) subclass of Bad_Accommodation F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 12 / 15
Case study: Religious tourism in Bari V Setting 12 positive examples (% of positive users’ feedback below 0 . 5) 39 negative examples (% of positive users’ feedback over 0 . 7) Experiment 3: restriction to only B&Bs hasAmenity some (Pets_Allowed) and hasAmenity some (Wheelchair_Access) subclass of Bad_Accommodation hasAmenity some (Bar) and hasAmenity some (Wheelchair_Access) subclass of Bad_Accommodation F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 13 / 15
Conclusions Application of ILP to SWS modeling for Integrated Tourism Learned axioms become part of the SWS descriptions Learned axioms are destination-specific Challenges: size and structure of data Ongoing work Improvement of the language bias specification for Foil - DL Experiments on main touristic destinations in Apulia Future work Learning from the feedback of specific user profiles Other case studies of Integrated Tourism outside Apulia F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 14 / 15
Questions? F.A. Lisi, F. Esposito (UNIBA) SWS Modeling by Learning from Users’ FeedbackILP 2014 - Sept. 14, 2014 15 / 15
Recommend
More recommend
