Relevance and Usage of Semantic Technologies in the

Factory of Things

Matthias Loskyll

Innovative Factory Systems (IFS)

German Research Center for Artiﬁcial Intelligence (DFKI)

Trippstadter Straße 122, D-67663 Kaiserslautern, Germany

Abstract. The Factory of Things (FoT) describes the extension of the Internet of

Things speciﬁc to the production domain. The semantic description of products,

processes and plants depicts a basic module of this approach, through which in-

formation can be ﬁltered and services can be discovered and orchestrated on de-

mand. The usage of semantic technologies in the context of production can make

a valuable contribution towards managing the growing complexity and increas-

ing the ﬂexibility and adaptability of production facilities. This position paper

discusses several use cases, which explain the potential advantages of using se-

mantic technologies in the ﬁeld of production automation. These scenarios cover

the interpretation of heterogeneous context data using ontologies, the semantic

description of facilities including the respective components (e.g. sensors), and

the orchestration of semantically annotated web services to build complex pro-

duction processes. Based on the described use cases, several scientiﬁc issues are

discussed with a special focus on semantic interoperability.

1 Introduction

The Internet of Things describes the ubiquitous networking of intelligent everyday ob-

jects, which communicate autonomously, exchange information and provide services.

The extension of this concept speciﬁc to the production domain is referred to as the

Factory of Things (FoT) [1]. The FoT includes the extended usage of ubiquitous infor-

mation and communication technologies in order to reach the networking of entities in

the production domain. Based on this networking, all types of information in a produc-

tion environment can be collected. The resulting information explosion, however, can

only be mastered using a context- and knowledge-based provision and processing of

data. Furthermore, the different mechatronic capabilities of a production plant, which

are encapsulated and represented as services, need to be orchestrated in order to deﬁne

complex production processes. Therefore, a semantic description of products, processes

and production plants is needed.

Semantic technologies allow the formal description of data and support the semantic

processing of data by machines, i.e. the interpretation of electronically stored pieces of

information with regard to their content and meaning. The formal, explicit representa-

tion of knowledge forms the cornerstone of the Semantic Web [2] and includes both

the modeling of knowledge and the deﬁnition of formal logics, which provide rules to

Loskyll M..

Relevance and Usage of Semantic Technologies in the Factory of Things.

DOI: 10.5220/0003352200640072

In Proceedings of the International Workshop on Semantic Interoperability (IWSI-2011), pages 64-72

ISBN: 978-989-8425-43-0

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

draw inferences over the modeled knowledge base. While several semantic modeling

approaches for the representation of knowledge with different expressional power exist

(e.g. taxonomies, thesauri, topic maps [3]), ontologies depict the most popular and the

most powerful approach of explicit knowledge representation. Ontologies, often deﬁned

as ”‘an explicit speciﬁcation of a conceptualization”’ [4], enable the modeling of infor-

mation as an independent knowledge base. Ontologies consist of three basic structures,

namely classes (or concepts), relations and instances. In addition, restrictions, rules and

axioms can be deﬁned in order to model complex coherences.

Based on respective description languages like OWL [5] and RDF [6], ontologies faci-

litate the structured exchange of information among heterogeneous systems, resulting

in a semantic interoperability. Another major advantage of ontologies is the possibility

to draw inferences over the explicitly modeled knowledge, thereby deriving new know-

ledge that is contained implicitly in the knowledge base. To this end, a reasoning system

is needed, which depicts a piece of software that is able to interpret logically deﬁned

facts and axioms and to infer logical consequences.

In the context of the FoT, semantic technologies are essential to ensure a knowledge-

based interpretation of information and to facilitate an efﬁcient orchestration of ser-

vices. This affects the representation of knowledge about products and plants, but also

the semantic description of services (Semantic Web Services). The latter one is neces-

sary because the description of a service’s interfaces and functionalities via standards

like WSDL [7] is of a mere syntactic nature. Similarly, deﬁning an orchestration of ser-

vices using languages like BPEL [8] happens in a syntactic manner. As a result, there

is a semantic gap between the syntactic description of web services and the underly-

ing meaning. On the basis of a semantic annotation, for example using technologies

like SAWSDL [9] or OWL-S [10], the meaning of a web service deﬁnition can be de-

scribed in a machine-understandablemanner. This additional semantic layer enables the

dynamic discovery of semantically described services and the (semi-)automatic orches-

tration of services to build higher-value services or even production processes.

This position paper presents our opinion about the importance of deploying semantic

technologies in smart factory environments and in a future Factory of Things. After a

discussion of related work in Section 2, this opinion is supported by several illustrating

usage scenarios, some of which have been implemented already in several experimental

setups (Section 3). Subsequent to the description of these scenarios, the most important

scientiﬁc issues that arise from the future usage of semantic technologies in the do-

main of industrial production are discussed (Section 4). In Section 5, we present our

conclusions and discuss opportunities for future work.

2 Related Work

While in several ﬁelds of application a central role is assigned to semantic technologies,

their usage has not gained acceptance in the ﬁeld of industrial production yet. The main

reasons for that might be the lack of illustrating application examples as well as the

complexity of the technical implementation in production environments.

Nevertheless, within the last years more and more research is carried out on the appli-

cation of semantic technologies to the production domain. Ontologies are frequently

used to build a common interaction model for agents operating the production process

[11] [12]. Several approaches have been made to deﬁne semantic plant and component

models in order to support a condition based maintenance of production plants [13] [14]

or ﬂexible reconﬁguration of assembly systems [15].

With the evolving usage of service oriented architectures in production systems, the se-

mantic description of services becomes more important. Particularly, within the scope

of the SOCRADES project, several concepts for the usage of ontologies in the pro-

duction domain [16] and for the semantic discovery and orchestration of services to

production processes [17] [18] have been developed.

These approaches are closely related to the ideas described in this position paper. How-

ever, we describe an industry-related usage scenario, which covers several issues that

arise from the special demands resulting from the usage of semantic technologies in the

production domain. This scenario will be implemented in a real-world research facility,

which is comparable with real manufacturing plants. Furthermore, we integrate several

forms of using semantic technologies in production, namely the semantic description

of products, processes and plants (overall structure, components, sensors), the semantic

interpretation of contextual information (e.g. location, sensor values, state of the plant)

and the dynamic discovery and orchestration of semantically described services. Be-

cause of the incorporation of these different semantically enriched subsystems based

on the formal representation of a vast amount of heterogeneous knowledge sources, our

approach is especially interesting for the research aiming at a semantic interoperability.

3 Area of Application and Use Cases

The beneﬁts of applying semantic technologies to the ﬁeld of production automation

are going to be demonstrated by means of different use cases in the SmartFactory

[21]. The SmartFactory

is the ﬁrst vendor-independent research and demonstration

facility for the application and evaluation of smart production technologies and includes

both research institutes and several partners from industry. It operates a hybrid, mod-

ular demonstration plant, which produces colored liquid soap, in a 200 square meters

industrial facility. Figure 1 shows a part of the modular soap production plant. In the

continuous ﬂow process, the transparent raw soap is heated and mixed with colorant

pigments depending on the customer order. The colored soap is ﬁlled into bottles, which

are then mounted with a dispenser, labeled and commissioned in the discrete produc-

tion part subsequently. Thereby, all relevant information about the product’s production

lifecycle is stored on an RFDI tag, which is located directly on the bottle.

One of the central research topics investigated in the scope of the SmartFactory

how a transition from function-oriented to service-oriented architectures (SoA) in pro-

duction can be achieved. To this end, a part of the plant has been converted to a service-

oriented control architecture based on WSDL and BPEL. However, as described in

Section 1, an additional semantic layer is needed in order to allow a dynamic discovery

and efﬁcient (semi-)automatic orchestration of services. Such a service discovery and

orchestration based on semantic technologies is planned to be implemented within the

scope of the production plant of the SmartFactory

. The semantic description of the

production process, the production plant and the corresponding components and de-

vices serves as the basis for this approach. A hierarchical service model is needed to

connect the different semantic representations: a production process needs several ser-

vices to perform certain tasks within the different process steps, while the components

and devices of the plant provide their functionality via services. The services them-

selves must be described semantically as well in order to make a dynamic discovery

and (semi-)automatic orchestration possible. Furthermore, the semantic description of

the process, the services and the plant, for instance by means of common ontologies,

facilitate a semantic interoperability and therefore the improved interaction among het-

erogeneous subsystems.

Fig.1. Part of the SmartFactory

production test bed.

Combining these semantic descriptions of subsystems in an intelligent factory environ-

ment with the acquisition and interpretation of context information, it is possible to

reach a cognitive plant behavior. It is important to clarify that by cognitive plant be-

havior we do not mean that the plant becomes an autonomous intelligent system which

makes decisions by itself. The plant should always be aware of its own state and should

be able to make suggestions how to solve a problem instead, thereby supporting the

engineer or maintenance worker. This means that the human plays an essential role in

our vision of a Factory of Things.

In the soap production scenario, we plan to develop the recognition of defective de-

vices and errors in the production process by interpreting the information provided by

an OPC UA [22] server, which collects the data from several programmable logic con-

trollers (PLC) in our system. Having recognized such an incidence, the plant should

be able to decide by performing a reasoning over the modeled knowledge whether the

product can still be manufactured. This includes the dynamic discovery of ﬁeld devices

(e.g. pumps) that provide a similar service as the defective device.

In times of ever shorter product lifecycles and an increasing demand for customized

product variants, the ﬂexible adaption of the production process becomes essential.

Within the scope of the SmartFactory

, we investigate how such a ﬂexible reconﬁg-

uration of the production process can be achieved with the help of semantic services.

Therefore,we plan to implement an experimental setup demonstrating a semi-automatic

orchestration, in which appropriate services are discovered by means of their semantic

description and the semantic speciﬁcation of the new product variant. An engineer can

then select the matching services for each production step. The deﬁnition of a new pro-

duction process can be improved signiﬁcantly by ﬁltering the contemplable services

using semantic templates instead of performing a pure syntactic search.

In order to gain further experience in the usage of semantic technologies in the ﬁeld

of production automation, we implemented several use cases, which cover different as-

pects of the complex scenario of soap production described above.

One demonstrative use case, for instance, deals with the topic of mobile maintenance

of production plants and of the corresponding ﬁeld devices. Thereby, a maintenance

worker is supported by a seamless navigation application (Figure 2) running on a mo-

bile device that guides the worker to the faulty ﬁeld device. To this end, the data of

several context sources (e.g. indoor positioning systems) is collected and interpreted.

By querying an ontology using this contextual information, explicit knowledge about

the present situation can be inferred. This information is then provided to the mobile

navigation device, which is able to adapt its user interface depending on the derived sit-

uation. In the scope of this usage scenario, we implemented a function library, based on

the OWL API 3.0 [19], which can be used to make queries against the respective ontol-

ogy and to draw inferences independent from the used ontology or reasoner. By the use

of this technology, the knowledge about the meaning of different context sources is not

hard coded in the source code of the application anymore, but it is explicitly modeled

in an extensible knowledge base. As a result, only the ontology needs to be altered if

the usage scenario or the environmental conditions (e.g. sensor setup) change.

Fig.2. Ontology-based location-aware maintenance application.

A second use case, which has been implemented in the form of an industry-related ex-

perimental setup as depicted in Figure 3, deals with the dynamic discovery of services

provided by ﬁeld devices using semantic annotations. In the process of this demonstra-

tor, pills are ﬁlled into bins according to the information stored on the RFID tag located

at the bin, i.e. the product itself carries the information about its production order. Af-

ter the ﬁlling process, a camera performs a quality control by counting the number of

pills ﬁlled to the bin using image recognition. The difference to the approach described

by Stephan et al. [20] is that the different ﬁeld devices included in the production pro-

cess (e.g. RFID reader devices, inductive and ultrasonic sensors, pneumatic stoppers,

camera) are not controlled by a programmable logic controller (PLC) anymore, but

they provide services based on WSDL. The orchestration of the services happens via

a BPEL engine running on a central server, which controls the production process. In

order to facilitate an efﬁcient discovery and binding of the services provided by the dif-

ferent ﬁeld devices, we annotated the corresponding WSDL ﬁles using SAWSDL. The

concepts that are referenced by the semantic annotations (e.g. device categories, ser-

vices, basic operations) are modeled in an OWL ontology. The ﬁeld devices subscribe

automatically using DPWS, which causes their SAWSDL ﬁles to be parsed using the

SAWSDL4J API and the resulting information about the provided services and oper-

ations to be stored as instances in the ontology. These instances are deleted from the

ontology as soon as the respective devices unsubscribe again.

Fig.3. Experimental setup: semantic services provided by ﬁeld devices.

4 Scientiﬁc Issues

Several scientiﬁc issues concerning the usage of semantic technologies in the produc-

tion domain arise from the conceptualization and implementation of the usage scenarios

presented in Section 3. The complexity of future intelligent factory environments goes

beyond the scope of previous ﬁelds of application of semantic technologies in most

cases. As a result, it is necessary to investigate which forms of semantic technologies

are qualiﬁed for the description of products, processes, services and plants. Is the ex-

pressiveness of today’s ontology languages like OWL sufﬁcient to model the complex,

heterogeneous knowledge or should we include additional techniques like rule systems?

How can semantically described services be orchestrated to complex production pro-

cesses in a dynamic manner? The application of semantic technologies to this domain

offers a great possibility to identify special requirements for the future improvement of

these technologies.

Closely linked to the topic of knowledge representation, the issue of knowledge acqui-

sition is commonly known to be one of the major constraints in the development of

knowledge based systems [23]. Thereby, the great challenge lies in identifying appro-

priate knowledge sources and in developing suitable methods to extract the contained

knowledge. Several approachesexist to break this ”‘knowledge acquisition bottleneck”’

[24] [25] [26]. However, in the domain of industrial production, the issues of knowledge

acquisition and knowledge update become even more difﬁcult because of the variety of

heterogeneous knowledge sources such as industrial standards, manuals, speciﬁcations,

stocklists or CAD models.

When discussing the usage scenarios described in Section 3, especially the complex

soap production process, it becomes apparent that a uniform interaction among the dif-

ferent subsystems like the service orchestration and the control of the facility is impor-

tant. To this end, common interoperability models are needed, which make the meaning

of data understandable for each subsystem. The usage of common semantic models

(e.g. an ontology describing the plant and the contained components like ﬁeld devices

or sensors, the semantic description of services, semantic device models) can help to

facilitate a semantic interoperability among heterogeneous systems. The central issue is

the speciﬁcation of a common vocabulary for the semantic description of different in-

dustrial systems in a vendor-independent manner. For that, commonly agreed standards

would have to be deﬁned. As long as such standards do not exist, advanced methodolo-

gies for ontology mapping [27] are needed.

In order to make the developed concepts, which deal with the usage of semantic tech-

nologies in smart factory environments, applicable to existing industrial systems, it is

necessary to examine how such systems can be migrated to new implementations and

architectures and how much resources it would take to build new factory systems that

incorporate semantic technologies. The discussion of these issues is essential for the

success of future research on the usage of semantic technologies in industrial produc-

tion.

5 Conclusions and Future Work

In this position paper we discussed the relevance of semantic technologies in the vision

of a future Factory of Things on the basis of several use cases. We identiﬁed the most

important scientiﬁc issues that arise from the usage of semantic technologies in the

domain of industrial production. Apart from the semantic description of products, pro-

cesses, services and plants, the knowledge acquisition problem depicts a cornerstone of

our approach. Furthermore, the usage of semantic technologies for the creation of a se-

mantic interoperability among heterogeneous systems in the production domain offers

a highly interesting area for future research. In this context, we are going to investigate

different mapping algorithms, but also possibilities to automatically extract synonyms

and similarity relations from thesauri like WordNet [28].

Further scientiﬁc issues will be addressed in our future research and in the implemen-

tation of an industry-related usage scenario within the scope of the SmartFactory

. In

the near future, we are going to evaluate the expressiveness requirements of our pro-

cesses based on the workﬂow patterns approach [29] because it is necessary to assess

whether standard modeling approaches like BPEL, BPML or OWL-S are qualiﬁed to

describe complex industrial production processes. Another important topic of future

work deals with uncertainty and fuzziness of information, especially sensor values, in

intelligent factory environments. To address this issue adequately, we have to consider

methods such as probabilistic ontologies [30] [31], ontologies extended by fuzzy de-

scription logics [32] or combination with Bayesian networks [33].

In order to make the implemented use cases applicable to industrial production, how-

ever, further important issues like practicability, security or real-time capability must be

examined. We believe that the usage of semantic technologies in the context of produc-

tion can make a valuable contribution towards managing the growing complexity and

increasing the ﬂexibility and adaptability of production facilities.

References

1. Z¨uhlke, D.: SmartFactory - Towards a factory-of-things. Annual Reviews in Control, Vol. 34,

Issue 1, 129–138, ISSN 1367-5788 (2010)

2. Berners-Lee, T., Fischetti, M.: Weaving the Web: The Past, Present and Future of the World

Wide Web by its Inventor. Britain: Orion Business (1999)

3. Garshol, L.M.: Metadata? thesauri? taxonomies? topic maps! making sense of it all. Journal

of Information Science, Vol. 30, No. 4, 378-391 (2004)

4. Gruber, T. An Translation Approach to Portable Ontology Speciﬁcations. Knowledge Acqui-

sition, Vol. 5, No. 2: 199–220 (1993)

5. Smith, M. K., Welty, C., McGuinness, D. L.: OWL Web Ontology Language Guide (2004)

6. Manola F. and Miller, E.: RDF Primer. W3C Recommendation (2004)

7. Christensen, E., Curbera, F., Meredith, G., Weerawarana, S.: Web Services Description Lan-

guage (WSDL) 1.1 (2001)

8. Alves, A., Arkin, A., Askary, S., Bloch, B., et al.: Web Service Business Process Execution

Language Version 2.0. OASIS (2006)

9. Farrell, J., Lausen, H.: Semantic annotations for WSDL and XML schema. Technical report,

W3C (2007)

10. Martin, D. et al.: Bringing semantics to web services with owl-s. World Wide Web 10, 243-

277 (2007)

11. Diep D., Alexakos C., Wagner T.: An Ontology-based Interoperability Framework for Dis-

tributed Manufacturing Control. 12th IEEE Conference on Emerging Technologies and Fac-

tory Automation (ETFA 07). IEEE Automation and Control System, Engineering manufac-

turing (2007)

12. Obitko, M. and Marik, V.: Ontologies for Multi-Agent Systems in Manufacturing Domain.

13th International Workshop on Database and Expert Systems Applications (DEXA ’02).

IEEE Computer Society, Washington, DC, USA, 597–602 (2002)

13. M. Viinikkala, S. Syrjl, S. Kuikka: A Maintenance Demand Analyzer - a Web Service based

on a Semantic Plant Model. 4th International IEEE Conference on Industrial Informatics

(INDIN 2006), Singapore, 16–18 (2006)

14. Jin, G., Xiang, Z., Lv, F.: Semantic integrated condition monitoring and maintenance of

complex system. 16th International Conference on Industrial Engineering and Engineering

Management, 670–674 (2009)

15. Lohse N., Hirani H., Ratchev S.: Equipment ontology for modular reconﬁgurable assembly

systems. International Journal of Flexible Manufacturing Systems. Vol. 17, No. 4, 301–314

(2005)

16. Lastra, M. und Delamer, I. M.: Ontologies for Production Automation. Advances in Web

Semantics I: Ontologies, Web Services and Applied Semantic Web. 276–289 (2009)

17. Lastra, M. und Delamer, I. M.: Semantic web services in factory automation: fundamental

insights and research roadmap. IEEE Trans. Ind. Informat., Vol. 2, No. 1, 1–11 (2006)

18. Cndido, G., Jammes, F., Barata, J., Colombo, A.: Semantic SOA approach to support agile

reengineering at device level. Proceedings of the 10th IFAC Workshop on Intelligent Manu-

facturing Systems (IMS-2010), Lisbon, Portugal (2010)

19. Horridge, M., Bechhofer, S.: The OWL API: A Java API for Working with OWL 2 Ontolo-

gies. CEUR Workshop Proceedings, Vol. 529 (2008)

20. Stephan, P., Meixner, G., Koessling, H., Floerchinger, F. and Ollinger, L. : Product-Mediated

Communication through Digital Object Memories in Heterogeneous Value Chains. IEEE

International Conference on Pervasive Computing and Communications (PerCom). 199–207

(2010)

21. Z¨uhlke, D. SmartFactory: From Vision to Reality in Factory Technologies. Proc. of the 17th

International Federation of Automatic Control (IFAC) World Congress, 82-89 (2008).

22. Mahnke, W., Leitner, S. H., Damm, M.: OPC Uniﬁed Architecture. Springer Verlag (2009)

23. Cullen, J. and Bryman, A.: The knowledge acquisition bottleneck: Time for reassessment?

Expert Systems, Vol. 5, No. 3, 216-225 (1988)

24. Li, Z., Yang, M. C. , Ramani, K.: A methodology for engineering ontology acquisition and

validation. Artiﬁcial Intelligence for Engineering Design, Analysis and Manufacturing Vol.

23, No. 1 (2009)

25. Welbank, M.: An overview of knowledge acquisition methods. Interacting with Computers,

Vol. 2, Issue 1, 83–91 (1990)

26. Mozina, M., Guid, M., Krivec, J., Sadikov, A., and Bratko, I.: Fighting Knowledge Acquisi-

tion Bottleneck with Argument Based Machine Learning. Proceeding of the 2008 Conference

on ECAI 2008: 18th European Conference on Artiﬁcial intelligence. Frontiers in Artiﬁcial

Intelligence and Applications, Vol. 178. IOS Press, Amsterdam, The Netherlands, 234–238

(2008)

27. Kalfoglou, Y. and Schorlemmer, M.: Ontology Mapping: The State of the Art. The Knowl-

edge Engineering Review, Vol. 18, 1–31 (2003)

28. Fellbaum, C.: A Semantic Network of English: The Mother of All WordNets. Computers

and the Humanities, Vol. 32 , Nr. 2 , 209–220 (1998)

29. Russell, N., ter Hofstede, A.H.M., van der Aalst, W.M.P., Mulyar, N.: Workﬂow Control-

Flow Patterns: A Revised View. BPM Center Report BPM-06-22 (2006)

30. Costa, P. C. G. and Laskey, K.B.: PR-OWL: A Framework for Probabilistic Ontologies.

2006 conference on Formal Ontology in Information Systems: Proceedings of the Fourth

International Conference (FOIS 2006), IOS Press, Amsterdam, The Netherlands, 237–249

(2006)

31. Ding, Z. and Peng, Y.: A Probabilistic Extension to Ontology Language OWL. 37th Hawaii

International Conference On System Sciences (HICSS-37) (2004)

32. Gao, M. and Liu, C.: Extending OWL by Fuzzy Description Logic. 17th IEEE International

Conference on Tools with Artiﬁcial Intelligence (ICTAI’05). IEEE Computer Society, Wash-

ington, DC, USA, 562–567 (2005)

33. Ding, Z., Peng, Y., Pan, R.: BayesOWL: Uncertainty modeling in Semantic Web ontologies.

Soft Computing in Ontologies and Semantic Web, Vol. 204 of Studies in Fuzziness and Soft

Computing, Springer (2005)