Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

SOFIA - Semantic Technologies and Techniques for Interoperable Information in Smart Environments. INDRA

868 views

Published on

  • Be the first to comment

SOFIA - Semantic Technologies and Techniques for Interoperable Information in Smart Environments. INDRA

  1. 1. Semantic technologies and techniques for interoperable information in smart environments Felipe Martín Anaya Jorge Juan Rodríguez Vázquez Technological Innovation Department Technological Innovation Department Indra Sistemas S.A. Indra Sistemas S.A. Madrid, Spain Madrid, Spain fmanaya@indra.es jjrodriguezv@indra.esAbstract—The current paper attempts to establish desirable experience and to enable the user to benefit from smartfeatures for software tools used in semantic webs. It would environments, [2] to promote innovation while maintainingallow to establish benchmarks to choose the optimal solution value of existing legacy, [3] to create new user interactionfor a specific situation, in this case the interoperability or skill and interface concepts and [4] to maintain cross-industryto deal with different environments. The SOFIA project, interoperability by creating an Interoperability Openhaving faced this situation, wants to share the chosen solution. Platform (IOP) as a platform for new services. IOP will Keywords-Interoperability, Ontology, Reasoner, SOFIA, Web foster innovation and will guarantee the future evolution ofsemantics smart environments based on embedded systems, both from a scientific/technological point of view and in terms of I. INTRODUCTION business. The project captures the specific aspects of smart spaces The SOFIA project is a three-year ARTEMIS project and combines the requirements for common solutions.started in January 2009 and involving twenty two partners Three applications or “verticals” are identified, whichfrom four different EU countries. SOFIA proposes an represent different kinds of space – in terms of scale,Internet-like revolution in physical space, aiming to make potential applications and services:“embedded information” in the physical world available for • Local personal spaces (e.g. car )smart services – connecting the physical world with the • Smart indoor spaces (e.g. home, office, university)information world and envisages that the way embedded • Smart city (e.g. extended infrastructure and facilitiessystems are used to construct will change, due to the like a subway station, shopping centre).ongoing digital convergence, not just only physical devicesbut also services. The key outcomes of the project relate to user interaction At the core of SOFIA project is the notion of smart paradigms for interacting in smart environments, theenvironment. A smart environment is an ecosystem of common interpretability solution between manyinteracting objects - e.g. sensors, devices, appliances and heterogeneous devices and embedded systems, and on theembedded systems in general - that have the capability to application development schemes that can mobilise newself-organize, to provide services and manipulate/publish developers for smart environments.complex data. Smart environments aspire to provideindividuals with a more satisfying experience in their II. RELATED WORKeveryday lives from every environment (home, work), There are several currently-available technologies thatfunctioning invisibly and unobtrusively in the background can be used to satisfy the semantic needs of the projectand freeing them from hazardous work, and tedious routine SOFIA. Each of these uses a different technique, covering atasks. different area in a different way. While some of them might The mission of SOFIA project is to create a semantic be more efficient in time consumption, others will have ainteroperability platform and selected set of vertical more strict consistency checking or will cover areasapplications to form an embedded system based smart neglected by others, e.g. some knowledge representationenvironment. The key factor in these smart environments languages.will be common, open information storage and search extent This paper attempts to offer an outlook of severalfor all embedded systems, regardless of their specific currently-available semantic technologies, trying toimplementation technology. In this vision, simple, local understand the differences between some of them. Suchmash-up applications will be built on open data and devices. intent will never be exhaustive, as the technologies are alive, The project targets encompass [1] to connect the real evolving as this is being written and resulting in unavoidablephysical world with the information world to enrich the user changes in the conclusions: We might discard one program, for instance, for not having a native implementation of RDF,
  2. 2. but maybe next week a new release of that same software the application had not been designed to support one of thewill include that feature. languages. Some technologies do support extended Notwithstanding this, the goal can be extended to analyze languages, but not natively, which reduces the efficiency:the interoperability of the technologies. SOFIA is partially While translation is not needed in this case, the tool hasbased in the following postulates, known as the Smart execution problems with components that it cannot modelEnvironment Axioms: [5]. • Interoperability is unavoidable step in the evolution of information world. IV. EVALUATION • There is no single technology that can master the An objective comparison would have to define the best variety of needs. practices and unambiguously test them, either in a numeric • World cannot be built or changed instantaneously. performance (for instance, measuring response time to a set • The cost of interoperability agreements increases input) or in a polar way (for instance, the yes-no question when moving towards implementations. "Does X support the language Y?"). This will allow to offer • A single company cannot build the world. a common benchmark, but even then it will not serve unambiguously to satisfy the needs of any specific project This is one of the challenges that SOFIA is to face. (i.e. SOFIA), as each project would be interested in different,Several technologies will be used in its environment and it sometimes weighed, features.will be up to SOFIA to deal with all of them, understanding García suggests that two of the most importantall the information that they can provide and, in turn, giving benchmarks should be the ability to import and export athem information and/or commands that they will given language, in his case RDF. A tool might support RDFunderstand. This paper will not consider exactly where is the either natively (such as Jena [6] or Sesame [7]) or non-border between the responsibilities and attributions of natively (such as Protégé or WebODE), and this difference isSOFIA, its Human users and its non-Human interlopers. not at all trivial in this case as has been explained before [8]. Another essential benchmark suite, supported by both Thus, we have researched several other papers, including García and Poslad [9], is the interoperability. This can beDr. Raúl García Castros thesis Benchmarking Semantic Web described as the ability of two systems to exchange, and use,technology (2008) and Prof. Stefan Poslads Ubiquitous information [10] or, in our field, the ability of twoComputing: Smart Devices, Environments and Interactions autonomously-developed software components to do so [11].(2009), among others. Each of those works has partially Interoperability, being clearly a need to exchange ontologies,covered of the interests that SOFIA has in semantic is also a problem of the semantic web [12].technologies. Several official websites have also been The cooperation of different components to reach aconsulted, including websites of products and of the World common goal must be achieved through a coordination ofWide Web Consortium. Please refer to the bibliography for a interactions using explicit communications, either throughcomplete list of consulted works. the use of a central coordinator (orchestration) or a distributed one (choreography). Unfortunately, there is not III. TECHNICAL APPROACH even a de facto standardization to specify which the best A semantic technology, as considered in this environment, solution is [13].will have to obtain information coded in a knowledge There are several factors affecting interoperability, being first and foremost the heterogeneity of the formalisms torepresentation language. There are many of these, with models the system information [14]. This heterogeneity canimportant differences: Some of them are based only on be found not only in the information level, but in the systemframes, e.g., and other use also description logics. level too [15]. Some of the most important ontology languages may be Another factor also affecting interoperability goesRDF [2] (Resource Description Framework), designed as a beyond the sign to reach the meaning of it: More thanmetadata data model; its extension RDFS (RDF Schema), occasionally, an ontology will not have an exact equivalentthat can describe ontologies; OWL [3] (Web Ontology for a meaning purported by a different ontology, which couldLanguage), the W3C de facto standard for life sciences; its lead to an important problem[16].several extensions OWL DL, OWL Full, OWL Lite and The architecture M3, used by SOFIA, is theOWL 2 [4]; XML Schema, etc. implementation of the architecture OIP developed by Nokia Other tools will use representation models beyond those and VTT. SOFIA follows the M3 principles that deal withspecifically designed for this subject, including Open interoperability in several points, specifically: [17]Biomedical Ontologies (OBO) or Unified ModelingLanguage (UML), which only add to the difficulties of • Interoperability agreements on information level:SOFIA. o Common ontology model and data One of the greatest problems when trying to apply a presentation format are the onlycommon layer to several ontologies comes when the current requirementstechnology cannot change from any ontology to another, be o Respecting the integrity and independenceit because of some imperfection in the software or because of devices
  3. 3. • Enabling cross-domain use cases: • Jena, an open-source semantic web framework that o Means and techniques must be use case supporting OWL. [25] independent • KAON2 Reasoner, whose infrastructure manages o Support for enforcement to device and OWL-DL, SWRL and F-Logic ontologies. [26] smart object manufacturers • Pellet Reasoner, an open-source OWL-DL reasoner. [27] Several attempts have been made to unambiguously • Racer Pro Reasoner, a closed-source semantic webcompare the existing technologies and techniques, but these reasoning system and information repository. [28]might be already obsolete due to the further changes on the • SweetRules, an integrated set of tools for semanticsoftware and, even if they are not, might not adjust to the web rules and ontologies. [29]specific needs of the current project. As such, while thecurrent paper will still analyze the results of previously- TABLE I.established benchmark suites, it will use its own Bossam FACT++ Hoolet Jena KAON2 Pellet Racer Sweet Rulesmethodology to measure the points that might be of interest TABLE II. support Incomplete Pro Notfor SOFIA In your paper title, if the words “that uses” can Expressivity OWL-DL ? Partial / under SROIQ(D) ? DL Incomplete OWL-DL SHIQ(D) SROIQ(D) SHIQ Applicableaccurately replace the word “using”, capitalize the “u”; if Entailment development Yes Yes model Incomplete Yes Yes Yes Nonot, keep using lower-cased. Consistency checking DIG Support ? No ? Yes Yes No OWL-DL model Yes ? Yes Yes Yes Yes Yes No No SPARQL Support No No No Yes Yes Yes No No V. CONCLUSIONS Rule Support Yes (SWRL, own) No Yes (SWRL) Yes (Own) Yes (SWRL) Yes (SWRL, others) Yes (SWRL) Yes (SWRL, others) Documentatio Not García [18] analyzed several semantic technologies in n available Yes Not readily readily Yes Not readily Yes Yes Non- Yesterms of importing and exporting languages, and centering in Free/ Free/ Free/ closed- Free/ open- open- Free/ open- Free/Closed closed- Free/ open- Licensing source source source source Source Several source sourcetheir approach to RDF. This feature favored the frameworkssupporting native RDF, such as the open-sourced Jena andSesame. The results also analyzed several other semantic a. SOFIAs Benchmark suite of reasoners.web tools that non-natively supported RDF, includingKAON [19] (which has been superseded by KAON2 since SOFIA initially attempted to determine which of thethe publication of this work) [20]; Protégé [21] (the most available technology was more suitable to its specific needs,extended editor); and WebODE. considering that the chosen solution could need some About imports, all the three failed when importing class adaptation to solve any important deficiency. We soonhierarchies with cycles; Protégé also failed when importing discovered that the defining features of a semanticinstances of multiple classes; and WebODE had several framework depend on the situation: There is not a single,problems with properties lacking domain and range. Both categorically best solution, not even with adaptation, andKAON and Protégé were unable to export classes or most are lacking in terms of interoperability.instances of multiple domains; Protégé also failed with As a consequence of this, the approach has changed.instances of multiple classes; and KAON had problems with Now, one of the mid-term goals of SOFIA is themultiple domains, undefined range or XML Schemes. development of a semantic environment allowing a highWebODE, the less extended of the trio, had no problems degree of interoperability – because SOFIA, by its nature,when exporting. will deal with a variety of technologies using a variety of Protégé was the only program that showed trouble in ontologies. Having identified some of the important featuresintegration, particularly with a class which was an instance that can define a good semantic framework, SOFIA isof multiple metaclasses; with data type properties without currently working in the adaptation and creation of this newdomain but with range; with data type properties whose technology.range was String; and with instances related to those. This E.g.: SOFIA uses three autonomous ontologies for eachapproach was made when trying to integrate Protégé with of its "verticals" (ie. Local personal spaces, smart indoorKAON. spaces and smart city); all of these are managed using When facing this problem, SOFIA approaches the Protégé. Each of the ontologies extends from a fourth one,question analyzing the different reasoners: called core ontology, which includes the common classes. • Bossam Reasoner, a RETE-based rule engine using This only adds to the interoperability of the system, as does native supports to reason over OWL and SWRL the use of the code generator. ontologies, and RuleML rules. The analysis was incomplete. [22] REFERENCES • FACT++, a reasoner implemented in C++ and supporting OWL-DL and theoretically OWL2 in the [1] Kindberg, T. & Fox, A. System software for ubiquitous computing. next release. [23] IEEE Pervasive Computing (2002) • Hoolet Reasoner, a reasoner that can deal with [2] http://ww.w3.org/RDF OWL-DL ontologies only after translating them. [3] http://www.w3.org/TR/owl-features/ [24] [4] Cuenca, B. et al. OWL2: The Next Step for OWL. Journal of Web Semantics, 2008.
  4. 4. [5] Corcho, O. A layered declarative approach to ontology translation [16] Euzenat ea: J. Euzenat, T. Le Bach, J. Barrasa, P. Bouquet, J. De Bo, with knowledge preservation. Frontiers in Artificial Intelligence and R. Dieng-Kuntz, M. Ehrig, M. Hauswirth, M. Jarrar, R. Lara, D. its Applications, IOS Press (2005). Maynard, A. Napoli, G. Stamou, H. Stuckenschmidt, P. Shvaiko, S.[6] http://jena.sourceforge.net/ Tessaris, S. Van Acker and Ilya Zaihrayeu. D2.2.2 State of the art on ontology alignment. Technical report, Knowledge Web, 2004.[7] http://www.openrdf.org/doc/sesame/users/ch01.html [17] SOFIA Technical Annex (Dec 2008), Grant agreement no.: 100017[8] García Castro, R. Benchmarking Semantic Web technology. Universidad Politécnica de Madrid. (2008) [18] García Castro, R. Benchmarking Semantic Web technology. Universidad Politécnica de Madrid. (2008)[9] Poslad, Stefan. Ubiquitous Computing: Smart Devices, Environments and Interactions. Wiley. 2009. [19] http://www.sourceforge.net/projects/kaon[10] Several authors. IEEE Standard Glossary of Software Engineering [20] http://kaon2.semanticweb.org/ terminology. IEEE, 1991 [21] http://protege.stanford.edu/overview/protege-owl.html[11] Duval, E. Learning technology standardization: making sense of it all. [22] http://bossam.wordpress.com/. International Journal on Computer Science and Information Systems. [23] http://code.google.com/p/factplusplus/ 2004 [24] http://owl.man.ac.uk/hoolet/.[12] Brachmann: R. Brachmann and H. Levesque. Readings in Knowledge [25] http://openjena.org/ Representation. Morgan Kaufmann, San Mateo, 1985. [26] http://kaon.semanticweb.org/[13] Poslad, Stefan. Ubiquitous Computing: Smart Devices, Environments and Interactions. Wiley. 2009. [27] http://pellet.owldl.com/[14] Brachmann: R. Brachmann and H. Levesque. Readings in Knowledge [28] http://www.racer-systems.com/products/racerpro/index.phtml Representation. Morgan Kaufmann, San Mateo, 1985. [29] http://sweetrules.projects.semwebcentral.org/[15] Sheth, A. Interoperating Geographic Information Systems. Kluwer, 1998.

×