An approach to improve semantics in Smart Spaces using reactive fuzzy rules Natalia Díaz Rodríguez & Johan Lilius Turku Centre for Computer Science (TUCS), Åbo Akademi University, Turku, Finland M.P. Cuéllar & Miguel Delgado Calvo-Flores University of Granada, Spain 1 IFSA World Congress - NAFIPS 2013, Edmonton, Canada 25.6.13
15
Embed
IFSA World Congress -NAFIPS 2013 Edmonton, Alberta. Natalia Díaz
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
An approach to improve semantics in
Smart Spaces using reactive fuzzy rules
Natalia Díaz Rodríguez & Johan Lilius Turku Centre for Computer Science (TUCS),
Åbo Akademi University, Turku, Finland M.P. Cuéllar & Miguel Delgado Calvo-Flores
University of Granada, Spain
1
IFSA World Congress - NAFIPS 2013, Edmonton, Canada 25.6.13
Introduction
§ Smart Spaces (UbiComp): interoperability, working on behalf of the user, handle unanticipated situations
§ Human behaviour modelling: crucial task in AmI environments
§ Problem: Sensor data as crisp events.
2
Introduction
§ Context-awareness for machine processing and reasoning.
Our proposal: A hybrid framework for context-aware Smart Space application development
2 main components: Crisp KB and Fuzzy KB, connected by a rule engine which handles each type of event subscriptions. a. CRISP element: SPARQL Subscription-based RDF store with
(semantics + fuzzy reasoning expressive power) views allow a versatile framework for developing Smart Space applications.
– strategy allows loosening of semantics or efficiency (depending on application needs): avoiding continuous querying for changes or fuzzy discretization-based solutions.
– Knowledge representation & queries are + flexible § fuzzyDL still does not support some required semantic constraints § 2 KBs = Redundancy (twofold updates) -> advantage for
optimizing execution time of different queries/ datasets
Future Directions
13
§ Rule serializers (for crisp and fuzzy architectures) § Scalability + performance experiments in real-life applications
(reasoning, querying/updating and subscription response VS ontology size & fuzzy/crisp/hybrid rules)
§ Optimization: Semantic generalization of hybrid antecedents 1. When strict semantics are to be preserved (we sacrifice
performance): – IF (WeatherSituationTurku, isCurrently, VeryStormy) –> IF (WeatherSituationTurku, isCurrently, ?).
§ Pub/sub fuzzy reasoner (->Fuzzy SPARQL?) § Alternatives to data redundancy keeping
consistency? § Overall: Standardization of fuzzy SPARQL extension for fuzzy reasoning will achieve more usable, flexible, personalized and adaptive Smart Spaces