Building Semantic Sensor Web: Knowledge and

Interoperability

Salvatore F. Pileggi, Carlos E. Palau and Manuel Esteve

Universidad Politécnica de Valencia, Camino de Vera S/N, 46022 Valencia, Spain

Abstract. Semantic Sensor Web would be an evolving extension of Sensor

Web that introduces a semantic layer in which semantics or meanings of infor-

mation are formally defined according to well-defined semantic schemas (On-

tology). Semantics should improve the capabilities of collecting, retrieving,

sharing, manipulating and analyzing sensor data (or associate phenomena) pro-

viding a new interoperability model: semantic interoperability introduces the

interpretation of means of data allowing the engineering of novel architectures

based on standard reasoners.

1 Introduction

The idea of sensor networks, disseminated everywhere around the world as part of

everyday life, implicitly assumes they are not connected between them as well as

associated information systems are not integrated. This scenario can be summarized

as a great amount of data but a poor knowledge.

The term Sensor Web was first used in 1997 [1] to describe a novel sensor system

model where individual and autonomous nodes could act and coordinate as a whole

performing stand alone observations or cooperative tasks. Sensor Web is commonly

defined as 'Web-accessible sensor networks and archived sensor data that can be

discovered and accessed using standard protocols and application interfaces'.

Sensor Web is a general purpose concept that is progressively assuming great im-

portance within several application domains: large scale geographic information sys-

tem (GIS), social sensors and all sensor systems working in accordance with complex

business models that assume the cooperation of remote services could be easily suited

within Sensor Web. Sensor Web is a progressive concept mainly limited, at the mo-

ment, by the lack of standardization of access infrastructures and data models as well

as by weak business models.

Semantic Sensor Web would be an evolving extension of Sensor Web that intro-

duces a semantic layer in which semantics or meanings of information are formally

defined. Semantics should integrate web-centric standard information infrastructures

improving the capabilities of collecting, retrieving, sharing, manipulating and analyz-

ing sensor data (or associate phenomena) as well as potential interoperability between

systems through semantic interactions [4], [5].

F. Pileggi S., E. Palau C. and Esteve M..

Building Semantic Sensor Web: Knowledge and Interoperability .

DOI: 10.5220/0003112000150022

In Proceedings of the International Workshop on Semantic Sensor Web (SSW-2010), pages 15-22

ISBN: 978-989-8425-33-1

 2010 SCITEPRESS (Science and Technology Publications, Lda.)

The paper proposes a discussion about some of most relevant open research issues

addressed by Semantic Sensor Web with special focus on interoperability and know-

ledge engineering. Also related challenges are briefly discussed. On the contrary, an

exhaustive analysis of last generation semantic technologies, even if interesting, is out

of paper scope.

The paper is structured in order to first provide an overview on Semantic Sensor

Web model. Then, the interoperability model and related issues (Semantic Actors and

Shared Vocabularies) are introduced. In the last part of the paper, semantic know-

ledge is analyzed with special focus on most common applications. Both resource-

centric (Domain Ontology) and data-centric (Data Ontology) semantic schemas are

considered.

2 Related Work

At the moment, semantic technologies are applied in several sensor architectures.

Common applications have the aim of provide advanced support to information de-

scription and processing [3], data management [6], interoperable networking [5],

dynamic representation of situations and system states [7], advanced analysis of data

[9] and classification [10].

Semantic Sensor Web [4] would be a generalized concept in which semantic tech-

nologies allow interoperable interchanging of semantic data. A semantic environment

for Sensor Web addresses several research issues and challenges. Probably, the engi-

neering of semantic knowledge is the most interesting for its central and key role as

well as for the fundamental lack of standardized methodologies.

3 Semantic Sensor Web

The reference semantic environment for this work is showed in Figure 1: physical

systems are provided with a semantic representation according to a well-defined se-

mantic schema (ontology). Generic Semantic Actors are able to semantically interact

among them and with any external system through the interchange of semantic data;

these actors are designed on the top of the technologic ground provided by standard

reasoners and, so, they can work within a semantic interoperable context.

A semantic reasoner, reasoning engine, rules engine, or simply a reasoner, is

commonly defined as ‘a piece of software able to infer logical consequences from a

set of asserted facts or axioms’. The notion of a semantic reasoner generalizes that of

an inference engine, by providing a richer set of mechanisms to work with. The infe-

rence rules are commonly specified by means of an ontology language, and often a

description language.

Several research issues are addressed by semantic environments for the various

application domains. In the context of this work, semantic interoperability (among

sensor systems) and semantic sensor knowledge engineering are considered.

Fig. 1. Schematic view of the Semantic Sensor Web model.

3.1 The Interoperability Model

Semantic interoperability would improve common interoperability models: basic

interoperability assumes the interchange of messages among system without any

interpretation; functional interoperability integrates basic interoperability model with

the ability of intepretating data context under the assumption of a shared schema for

data fields accessing; semantic interoperability introduces the interpretation of means

of data. Semantic interoperability is a concretely applicable interaction model under

the assumption of adopting rich data models (commonly called Ontology) composed

of concepts within a domain and the relationships between those concepts.

Semantic technologies are partially inverting the common view at actor intelli-

gence: intelligence is not implemented (only) by actors (that are understood as inter-

preters) but it is implicitly resident in the knowledge model. In other words, schemas

contains information and the “code” to interpretate it.

Semantic interoperability is based on the capability of interoperable actors (called

Semantic Actors) built on the top of standard reasoners and able to interpretate gener-

ic semantic schemas.

3.1.1 Semantic Actors

The availability of standard languages for ontology definition and specification (e.g.

Ontology Web Language, OWL [11]), as soon as extensible reasoners (e.g. Jena,

Pellet [8]) able to automatically process semantic structures, allows the engineering of

intelligent semantic actors.

As showed in Figure 2, ontologies defined according to standard and interoperable

technologies (e.g. RDF or OWL) are inputs for actors designed on the top of standard

reasoners able to automatically interpretate ontology semantics; extended intelligence

layers could be designed providing actors with domain-specific querying and/or add-

ed capabilities (e.g. learning or multi-ontology computation).

Ontology-driven interaction takes advantage by semantic rules and relationships

implemented by ontology: the actor receives in input the ontology, it is able to inter-

pretate it; so, it does not need to implement semantic rules and relationships (com-

monly in dependence of reference data model) that, now, are expressed in the schema

according to an interoperable model.

Several software layers can be built as extension of the model represented in Fig-

ure 2; these layers can be designed for reaching several goals; common solutions

provide extended functions for supporting ontology-aware computation: actors are

able to switch their behavior and functionalities in function of the input ontology.

Further extended functionalities are learning (actors are provided with a memory

that allow them to build knowledge on the base of their activities) and multi-ontology

computation.

Fig. 2. Semantic Actor model for Ontology-driven and Ontology-aware computation.

3.1.2 Shared Vocabularies

Semantic interoperability assumes information related with different systems poten-

tially represented according to different semantic schemas. This last aspect is one of

the critical issues in order to assure semantic interoperability.

At the same time, the intrinsic presence of multiples semantic schemas to represent

similar information could address several disadvantages: actors cannot be able to

understand when concepts from different ontologies are referring to the same seman-

tic concept; on the contrary, actors could assume that two concepts are referring the

same semantic concept just because they have the same name.

These and other similar cases could generate ambiguous situations that could gen-

erate errors or other undesired/unexpected results.

Fig. 3. Example of semantic linking among concepts using shared vocabularies.

If a strong level of interoperability is required, mentioned ambiguities could be

solved for example through context-aware interpretation of semantic concepts.

Shared vocabularies could represent a suitable solution: as showed in Figure 3,

concepts from different ontologies could be linked to concepts composing shared

vocabularies.

Several ambiguities derived by intrinsic heterogeneity of information could be

solved as well as both basic and advanced semantic computation could be favorably

affected.

The main disadvantage is the intrinsic difficulty of defining and standardizing

shared vocabularies.

3.2 Semantic Knowledge

Standardized methodologies for knowledge building [3] are a current open research

issue. Semantic Knowledge implicitly needs rich schemas that include structured

concepts, related properties as well as complex relationships among them.

Mapping real knowledge on semantic schemas is, probably, the most creative task

for the concrete engineering of Semantic Sensor Web.

Semantic environment could provide an interesting perspective for knowledge

building: first of all, semantic schema implies an “overall” concept of knowledge in

which concepts and relationships among them converge in the same structure; fur-

thermore, the semantic interoperability model allows knowledge building regardless

by any predefined schema.

Sensor domain proposes several peculiarities that allow simplest approaches than

for generic knowledge building. Even considering that each sensor system could have

its peculiarities that could be reflected on semantic representations, the majority of

systems could be represented according to two main semantic structures: the Domain

Ontology and the Data Ontology. The first one should support resource-centric inte-

raction; the second-one data-centric interaction.

A brief description of both Domain Ontology and Data Ontology is proposed in

the following sections.

3.2.1 Domain Ontology

The main goal of the Domain Ontology is the model of sensor domain. The really

extended sensor application domain advises the structuring of main domain in several

sub-domain as well as structured classification of resource according to different

perspectives.

In dependence of considered systems and applications, different aspects could

converge in the reference ontology.

In few simple words, the Domain Ontology has to describe the reference system:

in the majority of cases, not only physical resources are defined but also complex

relationships with them or other resources as well as the definition of abstract con-

cepts (e.g. logic resources).

Concrete implementations depend by interaction scope. Typically, the classifica-

tion of resources according to several perspectives (e.g. functional) is required.

The Domain Ontology has a key role for allowing search, discovery and all inte-

ractions that assume there is not previous information about the considered system.

Under the realistic assumption of complex knowledge resulting by semantic rela-

tionship among physical resources and external semantic concepts, a possible refer-

ence model is showed by Figure 4; as represented, the central concept (Physical Re-

source) is the result of the convergence between domain specific semantics and in-

ferred concepts representing relationships with external concepts (e.g. Network,

Host); an abstract semantic layer includes high level features definition and logic

resource; this last layer differs respect to lower layers because it does not define a full

sub-domain/domain: composing concepts have finite means only in the context of the

main domain (sensor domain).

Fig. 4. Structuring sensor domain knowledge.

In other words the main domain assumes a core sub-domain and an extended do-

main built on the top. The relationships with external domains can be simple associa-

tions in some cases or extremely complex relations in other cases.

Knowledge engineering methodology is currently an open research issue: building

knowledge on semantic ontologies could be relatively suitable considering interoper-

able schema and semantic annotation standards provided by current semantic tech-

nologies; there are several efforts oriented to promote a convergence among multiple

semantic domains (for example the use of standardized vocabularies for ontology

concepts[3]).

The reference model was built under the assumption that advanced information

systems and/or large scale web-centric applications work on abstract semantic con-

cepts (logic resource) and not only on basic concepts (physical resource).

3.2.2 Data Ontology

A great number of systems just need to interchange information, such as sensor data.

In this last case, systems require ad-hoc structures in order to search, discovery or

retrieve data. This class of semantic schema (Data Ontology) is conceptually different

respect to Domain Ontology (resource-centric structure) because data-centric.

The main purpose of a Data Ontology is defining the model for interchanging data

as well as the meaning of data.

Furthermore, semantic rules and relationships among the different fields and con-

cepts could allow several innovative scenario as well as complex models for manipu-

lation of data, intelligent filtering and other high level applications.

4 Conclusions

Semantic Sensor Web proposes an evolving extension of Sensor Web integrating

common models for knowledge representation with formal description of semantic or

meaning of information.

The most modern semantic technologies enable an innovative technologic envi-

ronment in which systems can interact among them interchanging semantic data in a

context of semantic interoperability. These novel interaction models allow the engi-

neering of advanced semantic actors built on standardized technologies as well as an

innovative vision of Sensor Web and related applications.

Regardless by concrete technologies or languages, Semantic Knowledge building

can be considered as the most creative and critical issue for the concrete realization of

Semantic Sensor Web. At the moment, it appears there are not concrete methodolo-

gies for knowledge building.

References

1. Mike Botts and Alex Robin. Bringing the sensor web together. Geosciences, pages 46–53,

1997.

2. Vincent Huang and Muhammad Kashif Javed. Semantic sensor information descrip-tion

and processing. In Second International Conference on Sensor Technologies and Applica-

tions, 2008.

3. Daniel Oberle et al., Dolce ergo sumo: On foundational and domain models in the smart-

web integrated ontology (swinto). Web Semantics: Science, Services and Agents on the

World Wide Web, 5(3), 2007.

4. Amit Sheth, Cory Henson, and Satya S. Sahoo. Semantic sensor web. IEEE Internet Com-

puting, 12(4):78–83, 2008.

5. Wang Ke, Salma Abu Ayyash, Thomas Little, Semantic Internetworking of Sensor Sys-

tems, 2004 IEEE International Conference on Mobile Ad-hoc and Sensor Systems.

6. M.Lqbal, H.B.Lim, W.Wang,Y.Yao, A Service-Oriented Model for Semantics-based Data

Management in Wireless Sensor Networks, 2009 International Conference on Advanced

Information networking and Applications Workshops.

7. Thirunarayan, K.; Henson, C.A.; Sheth, A.P.; , Situation awareness via abductive reasoning

from Semantic Sensor data: A preliminary report, International Symposium on Collabora-

tive Technologies and Systems, 2009.

8. Evren Sirin, Bijan Parsia, Bernardo Cuenca Grau, Aditya Kalyanpur, Yarden Katz, Pellet:

A practical OWL-DL reasoner, Web Semantics: Science, Services and Agents on the

World Wide Web , Volume 5 Issue 2, 2007.

9. Kawakami, T.; Bich Lam Ngoc Ly; Takeuchi, S.; Teranishi, Y.; Harumoto, K.; Nishio, S.;

Distributed Sensor Information Management Architecture Based on Semantic Analysis of

Sensing Data, International Symposium on Applications and the Internet, 2008.

10. Majavu, W.; van Zyl, T.; Marwala, T.; Classification of web resident sensor resources

using Latent Semantic Indexing and ontologies, IEEE International Conference on Sys-

tems, Man and Cybernetics, 2008.

11. OWL Web Ontology Language Overview, http://www.w3.org/TR/owl-features/.