ENABLING SEMANTIC ECOSYSTEMS AMONG

HETEROGENEOUS COGNITIVE NETWORKS

Salvatore F. Pileggi, Carlos Fernandez-Llatas and Vicente Traver

TSB-ITACA, Universidad Politécnica de Valencia, Valencia, Spain

Keywords: Cognitive Networks, Semantic Technologies, Distributed Computing, Sensor Networks.

Abstract: Cognitive Networks working on large scale are object of an increasing interest by both the scientific and the

commercial point of view in the context of several environments and domains. The natural convergence

point for these heterogeneous disciplines is the need of a strong advanced technologic support that enables

the generation of distributed observations on large scale as well as the intelligent process of obtained

information. An approach based on the Semantic Sensor Web could be the key issue for enabling semantic

ecosystems among heterogeneous Cognitive Networks.

1 INTRODUCTION

Cognitive Networks (Thomas, 2005) working on

large scale are object of an increasing interest by

both the scientific and the commercial point of view

in the context of several environments and domains.

In fact, during the last years, the research

activities about local phenomena and their

correspondent impact on global phenomena have

been object of great interest inside the scientific

community as well as in the context of public and

private research institutions. Concrete environments

could depend by the research scope and they can

significantly vary for size, amount and kind of

information, involved actors, etc. An ideal scenario

in this sense is a metropolitan area that provides a

complex heterogeneous ecosystem in which humans,

machines and the environment are constantly

interacting.

Common research activities at metropolitan area

level are mainly focused on the study of climatic or

environmental (e.g. chemical or natural element

presence or concentration) phenomena and of human

behaviour (behavioural patterns, traffic, noise, etc.).

The study of these phenomena, first of all,

interests the citizens (or concrete collectives)

because it can be a complex and exhaustive

feedback in order to improve the quality of life or to

provide specific services for the interested

collectives (allergic people for example). The

evolution of these phenomena in the medium and

large period, as well as its social impact, is object of

great interest in the context of different domains and

disciplines.

The natural convergence point for these

heterogeneous disciplines is the need of a strong

advanced technologic support that enables the

generation of distributed observations on large scale

as well as the intelligent process of the obtained

information.

Existent solutions at level of metropolitan area

are mainly limited by the use of obsolete/static

coverage models as well as by a fundamental lack of

flexibility respect to the dynamic features of the

most modern virtual organizations. Furthermore, the

centralized view at the systems is a strong limitation

for dynamic data processing and knowledge

building. Finally, the heterogeneous nature of data

and sources implies complex model for data

representation and the related knowledge has to be

analyzed according to several perspectives (e.g.

local knowledge, domain, cross-domain).

This paper would exhaustively discuss the

impact of the application of semantic technologies to

high scale cognitive network, enabling semantic

ecosystems among heterogeneous structures and

information.

The paper is logically structured in two main

parts. The first part has mainly the goal to define and

characterize semantic ecosystems in relation with

real environments. It also deals the main limitations

currently existing for the massive dissemination of

cognitive networks working on metropolitan scale.

487

F. Pileggi S., Fernandez-Llatas C. and Traver V..

ENABLING SEMANTIC ECOSYSTEMS AMONG HETEROGENEOUS COGNITIVE NETWORKS.

DOI: 10.5220/0003696004870492

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (SSW-2011), pages 487-492

ISBN: 978-989-8425-80-5

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

A short overview about the most innovative

solutions for each one of the key technologic aspects

featuring cognitive networks will be provided as

well as a short analysis about related business

models. Finally, in the last section, the impact of the

semantic technologies application is analyzed, as

key factor for the improving of the interoperability

level among heterogeneous networks, sub-networks

and data. Furthermore, the capabilities of knowledge

building and intelligent analysis of data can be

strongly improved.

2 COGNITIVE NETWORKS:

FROM SCIENCE TO REALITY

In this section, first the semantic ecosystems will be

defined and characterized both with their

relationships with real environments. Later, the

current approaches to concrete solutions and related

limitations are analyzed. Therefore, the main section

scope is the definition of a generic reference

scenario for semantic ecosystems in relation with

their technologic and economic sustainability. As it

will be discussed, there is, at the moment, a

significant gap between theoretical models and their

concrete application.

2.1 Semantic Ecosystems among

heterogeneous Cognitive Networks

In the context of this work, a metropolitan (or urban

(Wikipedia, Urban Ecosystem) ecosystem (Figure 1)

is defined as a large scale ecosystem composed of

the environment, humans and other living

organisms, and any structure/infrastructure or object

physically located in the reference area.

An exhaustive analysis of environmental and

social phenomena is out of paper scope. Just

considering that we are living in an increasingly

urbanized world. From recent studies, it appears that

this tendency will be probably followed also in the

next future. It is a commonly accepted assumption

that further increases in size and rates of growth of

cities will no doubt stress already impacted

environments as well as the social aspect of the

problem.

Considering this tendency is hard to be

controlled or modified, there are a great number of

interdisciplinary initiatives, studies and researches

aimed to understand the current impact of the

phenomena as well as to foresee the evolution of it.

These studies have, evidently, a scientific focus,

but they also could be of interest in the context of

the everyday life. In fact, modern cities change their

structure and physiology in function of human

activities that constantly act as inputs for the

feedback system. It is easy to imagine the great

number of services that could improve the quality of

life of citizens (or collectives) with a deep

knowledge of the environment.

As mentioned, the study of the human activities,

of the environmental and climatic phenomena is

object of interest in the context of several disciplines

and applications. All these studies are normally

independent initiatives, logically separated

researches and, in the majority of the cases, results

are hard to be directly related. This could appear a

paradox: interest phenomena happen in the same

physical ecosystem, involving the same actors but

the definition of the dependencies/relationships

among atomic results are omitted even if they are

probably the most relevant results.

The common point is the need of great amounts

of heterogeneous data, normally generated on large

scale (Akyildiz, 2002). They can be “simple”

measurements or complex phenomena, sometimes

hard to be detected. This overall approach according

heterogeneous model has a strong impact especially

in the representation and processing of the

information.

Summarizing, at now urban ecosystems have a

directly equivalent logic concept by a knowledge

perspective but its realizations are mainly

knowledge environments than effective knowledge

ecosystems.

2.2 Current Approaches and

Limitations

The normal technologic support for enabling

knowledge environment is the cognitive network

(Thomas, 2005) that assumes a physical

infrastructure (sensors) able to detect interest

information or phenomena and a logic infrastructure

able to process the sensor data (knowledge building)

eventually performing actions, responses or complex

analysis.

The parameters that can potentially affect the

“quality” of the applications or studies are mainly

the sensor technology (constantly increasing in

terms of reliability, precision and capabilities), the

coverage area, the amount of data and, finally, the

process capabilities.

Current solutions are hard to be proposed on

large scale due to the current limitations of the

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

488

massive sensors deployment on large scale (Pileggi,

2011c). Furthermore, the following limitations can

be clearly identified:

• Lack of social view at the information (

Rahman,

2010). Applications and studies have a

fundamental lack of interaction and cooperation,

even if they are part the same logic and physical

context. A social approach could increase the

possibilities of data sharing and collaboration.

• Static coverage models. High coverage areas

imply the need of sustainable infrastructures.

The common models that assume static nodes

can be a high expensive solution in a context of

high density of sensors. Mobile sensor networks

(Pileggi, 2009) could be a suitable solution for

environments, as metropolitan areas,

characterized by the presence of a great number

of mobile actors (e.g. humans, vehicles, bikes,

etc.).

• Obsolete view at resources. Physical and logic

networks are undistinguished with the

consequent lack of flexibility in distributed

environments. This is a strong limitation for a

great number of business scenarios as well as a

technologic restriction to the resource (physical

in this case) sharing among structured virtual

organizations (Foster, 2008) (Mell, 2009)

(Foster, 2001) (Pileggi, 2011c).

• Not always effective business models. Due to

the static view at resources and applications,

innovative scenarios are hard to be realized and

common business actors are hard to be

identified in real contexts.

The impact of the proposed points can be limited

if a distributed perspective for infrastructures and

information is assumed. Due to the heterogeneous

features of the data source and information, the

interoperability plays a key role for the effective

realization of the model.

In the next section, a distributed approach for the

main infrastructure is described both with the most

advanced solutions based on semantic

interoperability that allow a social perspective for

the knowledge. Also the analysis for the knowledge

building process based on the application of the last

generation contextual semantic is proposed.

3 THE IMPACT OF SEMANTIC

TECHNOLOGIES:

DISTRIBUTED APPROACH

The previous section propose an abstract model for

semantic ecosystems as a possible evolution of

cognitive networks to a distributed approach that

Figure 1: Logical overview at Semantic Ecosystems.

ENABLING SEMANTIC ECOSYSTEMS AMONG HETEROGENEOUS COGNITIVE NETWORKS

489

should allow the enablement of complex logic

ecosystems in a context of flexibility and economic

sustainability.

This conclusion is mainly motivated by the

objective difficulty of modelling virtual

organizations using centralized models as well as by

the low level of interoperability that currently

characterizes heterogeneous systems.

Distributed solutions objectively improve the

flexibility of architecture but they require a high

level of interoperability among systems especially if

they are not part of the same social and economic

context.

This section would discuss the benefits

introduced by semantic technologies as general

solution for improving the interoperability and as

key support for the processing of heterogeneous data

(knowledge building).

3.1 Semantic Interoperability

Considering a distributed sensor domain, the key

issue is the evolution of the Sensor Web model to

the Semantic Sensor Web that, in practice, assumes

systems interchanging semantic information on the

top of the common functional interoperable layer

(Pileggi 2010).

The current semantic model for the web is

affected by several problems. These open issues,

such as ambiguities and performances, are object of

an intense research activity that is proposing several

solutions as simplification or particularization of the

main model.

In order to enable effective working systems on

large scale, a simplified model of the semantic web

is considered (Pileggi, 2011b). It assumes semantic

reasoners operating over three interrelated semantic

structures (Figure 2):

• Ontology as in common semantic environment,

it has to represent data and knowledge at

different levels.

• Shared Vocabulary. It could be a contextual

structure that represents an “agreement” in order

to avoid possible ambiguities and semantic

inconsistencies inside semantic ecosystems.

• Semantic Link. Additional structures that should

link concepts from different ontologies and

concepts from vocabularies. These structures

can directly relate concepts from different

ontologies and they can indirectly build

contextual semantic environments.

As showed in Figure 2, the Ontology is a

semantic structure normally associated to a local

knowledge environment. Concepts from different

ontologies can be related at domain level through

semantic links to vocabularies concepts.

In the example represented in Figure 2, the

concepts c1 and c6 are equivalent to the concept c3

at domain level and so, at this level, they are also

equivalent to each other.

This schema could be an exhaustive model for

the great part of logic environment associated to a

concrete domain. But the heterogeneous features of

semantic ecosystems force the knowledge

environment to work in a multi-domain context.

This last aspect need a further semantic layer (Figure

2): Cross-domain Vocabularies are defined in order

to relate concepts from different domains through

semantic links.

In the example of Figure 2, the concepts c3 and

c4 are equivalent to c5 at global level. This also

implies that c1 (linked to c3) and c2 (linked to c4)

are equivalent to c5 and that they are equivalent to

each other at global level.

A short analysis of the model proposed in Figure

2 first of all puts in evidence the hierarchical

structure of the semantic knowledge building

according to an increasing level of abstraction.

On the other hand, the semantic model is

completely open and assures, through semantic links

to higher concepts, a high level of expressivity and

interoperability without forcing standard data

models.

This last aspect has a critical importance at

application level where models, rules and

relationships need integrations, particularizations

and extensions in function of concrete applications

and domains.

3.2 Knowledge Building

This second support is the natural complement to the

first one in order to provide systems with the

capability of building abstracted knowledge on the

base of basic sensor data on the model of (Pileggi,

2011a).

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

490

Figure 2: Interoperability Model Schema.

The main challenge is the generalization of this

approach on large scale and considering

heterogeneous environment. As showed in Figure 3,

a local knowledge schema (Ontology) is composed

of two kinds of concepts:

• Low-level Concepts: They have a mean only in

the context of their local knowledge

environment. The main consequence is the lack

of any class of semantic link. In practice, they

are low-abstracted information that normally is

“visible” only in the local system.

• High-level Concepts: They are a set of concepts

that naturally complete the previous one. In fact,

they are high-abstracted concepts that have

evidently a local mean but also a domain and/or

global mean.

A deeper analysis of the structure (Figure 3 on

the right) allows the definition of semantic layers

inside the main structure:

• Data Source. Set of low-level concepts that

represent the data-sources (sensors or any other

kind of physic/human data source).

• Data. As the previous one but representing data.

• Core. Abstracted layer composed of semantic

rules that relate low-level and high-level

concepts. Due to its critical role, this is the key

layer in the semantic structure.

• Domain-specific Layers. Any set of high-level

concept required in the context of concrete

domains and applications.

The main advantage introduced by the schema is

the possibility to have a common ground for data

source and data representation, as well as a clearly

defined set of standardized high level abstracted

concepts. Also the core part of the ontology, that has

the goal of building the knowledge of basic data, is

an ad-hoc component of specific applications. In the

context of an ideal semantic ecosystem, any class of

information (basic data or abstracted knowledge)

can be correctly interpretated in the context of the

owner system as well as inside other systems

socially connected.

4 CONCLUSIONS

The power of collecting and relating heterogeneous

data from distributed source is the real engine of

high-scale cognitive networks.

The economic sustainability, as well as the social

focus on the great part of the applications,

determines the need of an innovative view at

networks and architectures on the model of most

modern virtual organizations. These solutions

require a high level of interoperability, at both

functional and semantic level. The current

ENABLING SEMANTIC ECOSYSTEMS AMONG HETEROGENEOUS COGNITIVE NETWORKS

491

Figure 3: Local Knowledge Model.

“Semantic Sensor Web” approach assures a rich and

dynamic technologic environment in which

heterogeneous data from distributed source can be

related, merged and analyzed as part of a unique

knowledge ecosystem.

REFERENCES

Thomas (2005), R. W.; DaSilva, L. A.; MacKenzie, A. B.;

"Cognitive networks," 2005 First IEEE International

Symposium on New Frontiers in Dynamic Spectrum

Access Networks, 2005. DySPAN 2005. 8-11 Nov.

2005.

Akyildiz (2002), I. F.; Weilian Su; Sankarasubramaniam,

Y.; Cayirci, E.;, A survey on Sensor Network, IEEE

Communication Magazine, August, 2002, Volume 40,

Issue 8, Aug. 2002 Page(s):102 - 114.

I. Foster (2008), Yong Zhao, Ioan Raicu, Shiyong Lu,

Cloud Computing and Grid Computing 360-Degree

Compared, Grid Computing Environments Workshop,

2008. GCE '08.

P. Mell (2009) and T. Grance, Draft NIST Working

Definition of Cloud Computing v14, Nat. Inst.

Standards Technol., 2009. [Online]. Available:

http://csrc.nist.gov/groups/SNS/cloud-

computing/index.html.

Salvatore F. Pileggi (2011a), A Semantic Environment for

Data Processing in Embedded Sensor Networks,

Proceedings of the International Workshop on

Semantic Interoperability (IWSI2011), January 2011,

Rome, Italy.

Salvatore F. Pileggi (2011b), Carlos Fernandez-Llatas,

Vicente Traver, A Semantic Layer for Embedded

Sensor Networks, ARPN Journal of Systems and

Software, Vol.1 No.3, pp. 101-107, June 2011.

Salvatore F. Pileggi (2010), Carlos E. Palau, Manuel

Esteve, Building Semantic Sensor Web: Knowledge

and Interoperability, Proceedings of the International

Workshop on Semsntic Sensor Web (SSW2010),

October 2010, Valencia, Spain.

Urban ecosystem, Wikipedia, http://en.wikipedia.org/wiki/

Urban_ecosystem.

Salvatore F. Pileggi (2011c), Carlos E. Palau, Manuel

Esteve, Enabling Wireless Sensor Network within

Virtual Organizations, Future Internet Services and

Service Architectures, 2011, Eds. A. Prasad, J. Buford,

V. Gurbani, River Publishers.

Rahman, M. A. (2010); El Saddik, A.; Gueaieb, W.;

Building Dynamic Social Network From Sensory Data

Feed, IEEE Transactions on Instrumentation and

Measurement, vol.59, no.5, pp.1327-1341, May 2010.

Salvatore F. Pileggi (2009), A multi-domain framework

for Wireless Vehicular Sensor Network, Proceedings

of International Conference on Ultra Modern

Telecommunications and Workshops (ICUMT2009),

October 12-14, 2009, St. Petersburg, Russia.

Foster (2001), C. Kesselman, S. Tuecke, The Anatomy of

the Grid: Enabling Scalable Virtual Organizations,

International J. Supercomputer Applications, 15(3),

2001.

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

492