Naumenko A., Nikitin S., Zharko A.

Transcription

1 Int. J., Vol. x, No. x, xxxx 1 Agent-driven Semantic Web-based EAI Naumenko A., Nikitin S., Zharko A. Mathematical Information Technology Department, University of Jyväskylä Agora, P.O.Box 35, FIN University of Jyväskylä, FINLAND anzharko@cc.jyu.fi 1 Introduction At the current stage of ICT development, there is a diversity of heterogeneous systems, applications, standards of data representation and ways of interaction. All those systems were tailored for particular tasks and goals. The world is heterogeneous and we face the challenge by trying to integrate heterogeneous systems into a unified environment. However, today s state of affairs shows us improvements of data processing and acquisition from one hand, from another hand it is still difficult to process data by intelligent software that allows integration of heterogeneous systems. Data, represented in systems are in their own format, have no semantic description, and are often non-interoperable. Taking into account great variety of possible resource types, data formats and ways of accessing and acquisition, integration of such resources into unified environment is an important development challenge [2, 3]. Basically, the integration tasks can be studied from two perspectives: adaptation of heterogeneous applications and adaptation of heterogeneous data originally represented in different formats. Figure 1 depicts different types of data, which need to be integrated such as flat files, XML-based data or data from specific applications in specific formats, often non-interoperable. Copyright 200x Inderscience Enterprises Ltd.

2 Author Figure 1 - Data integration illustrated (adopted from [1]) The integration process may include the following key functions [4, 5]: - Extracting, transformation and loading for building data warehouse or operation data stores and giving end-user/sources/applications possibility to proceed integrated data; - Data replication, to allow multiple heterogeneous servers and databases to share data in real time; - Data Synchronization to allow the sharing of data between servers and remote devices when connectivity is temporary. Application adaptation another part of the general integration task. The data is generated by different resources with specific applications. Considering this part of integration we can distinguish following application specific features: - Application functions; - Application APIs; - Application interfaces. All variations of these features have effect on process of adaptation and architecture of adaptation framework. 2 EAI and Web Services The history of harnessing the potential of Internet by industrial companies includes 3 phases [6]: provision of information: companies utilize Internet to provide a static information about their activities;

3 Title information exchange: companies utilize Internet to perform a dialog with their clients; joint processes: companies use Internet for integration of their businessprocesses via web services. Very few business applications can live in isolation. Most often, applications have to be integrated with other applications inside and outside the enterprise. This integration is usually achieved through the use of some form of "middleware". Middleware provides the "plumbing" such as data transport, data transformation, routing etc. [7]. Enterprise integration can also help to reduce costs, increase operational efficiencies, expedite time to market, and improve return on information technology investments. Without enterprise integration, various infrastructures will lack the robustness and flexibility in a dynamic economy [8]. Now integration is the direction of success for enterprises. They invest large portions of their ICT budget on informational integration. As the technology of Web services becomes more and more popular, World Wide Web Consortium 1 (W3C) performing its task of leading the Web to its full potential has published a specification of Web Services Architecture [9]. They formulated the definition of Web service as a software system designed to support interoperable machine-to-machine interaction over a network. It has an interface described in a machine-processable format (specifically WSDL). Other systems interact with the Web service in a manner prescribed by its description using SOAP-messages, typically conveyed using HTTP with an XML serialization in conjunction with other Web-related standards [9]. A standardized Web service platform must implement the following stack (see also Figure 2): 1

4 Author SOAP (Simple Object Access Protocol): lightweight messaging protocol, based on XML [10]. It is aimed at structured information exchange between applications in a decentralized, distributed environment. WSDL (Web Service Description Language): it allows describing relations and interaction between Web services [11]. UDDI (Universal Description, Discovery and Integration): this standard includes a set of protocols dedicated to provision of public catalogues for real-time registration and search of Web services and also for other business processes [12]. Examples of Web service architecture are: E-Speak [13], ebxml 1, RosettaNet 2. Figure 2 - Web Services computing stack (adopted from [14]) Attempts to standardize the field of automated Web services composition have resulted in WSFL Web Services Flow Language [15] (see also Figure 2)

5 Title The evolution of technologies in industry also leads to Web services, and maintenance domain can be considered as an example. Now maintenance is based on online instrumentation monitoring, which assumes keeping track of real time operating conditions of industrial assets to aid [16]: operational performance; costs optimization; optimization of the instrumentation life expectancy; minimization of intervention in operational performance; Analysis of such approach in industry proves its utility [16] and it has strongly fixed in industry. The data about the state of manufacturing instrumentation (field devices) is provided by embedded analytical block (software + hardware). The latter can collect the data about the state of correspondent instrumentation and provide access to it from mobile or stationary terminals [17]. The field devices of new generation smart devices - usually can notice their fault states and notify responsible personnel via mobile phone, , etc. It is not always economically efficient to equip all field devices with such capabilities thus only crucial field devices can be enhanced in this way [16]. 3 Semantic Web Services The development of Web service technology and its standardization have encountered significant difficulties on their way. The main cause for that is that the process of automated Web services integration needs mature description languages for its support and also efforts of terminology standardization to provide interoperability between heterogeneous Web services. The facilities of

6 Author Web service search are limited; one can search by keywords, but cannot ask for something similar since UDDI does not provide a vocabulary (ontology) [18]. From this moment technology of Web services have started cooperative research with Semantic Web initiative. The Semantic Web is an initiative of the W3C, with the goal of extending the current Web to facilitate Web automation, universally accessible content, and the 'Web of Trust'. Current trends on Web development leading to a more sophisticated architecture are: Semantic Web, Device independence, Web Services. Tim Berners-Lee [19] has a vision of a semantic web, which has machine-understandable semantics of information, and trillions of specialised reasoning services that provide support in automated task achievement based on the accessible information. This gives a completely new perspective for the knowledge acquisition, knowledge engineering, and knowledge representation communities. Semantic Web is the vision of having data on the Web defined and linked in a way that it can be used by machines not just for display purposes, but for automation, integration and reuse of data across various applications. This vision assumes annotating World Wide Web artefacts with machine-interpretable descriptions of their underlying semantics, and provides mechanisms for automated reasoning about them. On the technology side, Web-enabled languages and technologies are being developed (e.g. RDF-Schema [20], DAML+OIL [21], OWL [23], DAML-S [22], see Figure 3), schema and ontology integration techniques are being examined and refined. The success of the Semantic Web will depend on a widespread adoption of these technologies.

7 Title Figure 3 - Semantic Web Services computing stack (adopted from [14]) Management of resources in Semantic Web is impossible without the use of ontologies, which can be considered as high-level metadata about semantics of Web data and knowledge. Ontologies are content theories about the sorts of objects, properties of objects, and relations between objects that are possible in a specified domain of knowledge. Ontological analysis not only clarifies the structure of knowledge but also enables knowledge sharing and integration. Ontologies to be shared need a standard representation and exchange language. DARPA 1 and W3C joined forces to make the DAML+OIL language (later converted to OWL). It builds on earlier W3C standards such as RDF and RDF Schema. DAML extends RDF and RDFS with richer modelling primitives. DAML-S (DAML for Services) provides an upper ontology for describing properties and capabilities of (Web) services in an unambiguous, computer interpretable markup language, which enables automation of service use by agents and reasoning about service properties and capabilities. Industrial enterprises understand that their successful integration cannot be performed without providing interoperability of information systems being in 1

8 Author use. On the symposium held on December 11, 2001 by the Silicon Valley World Internet Center [24], challenges that have to be addressed applying Semantic Web were discussed too. Since Semantic Web provides interoperability via common inter-enterprise taxonomies (ontologies), a challenge would be to join the efforts of enterprises to develop them. Ontologies represent the entire set of relationships within an enterprise [25]. The development of ontologies must be performed in a coordinated manner to ensure the consistency of the whole model. Taxonomies are defined by the participants as the vocabulary used inside the ontology. Language and terms need to be common among the users for the data to pass through the various relationships of the ontology and to mean the same thing to all parties. Successful applications of ontologies in decentralized and distributed environments require substantial support for change management in ontologies and ontology evolution. Among the most important European efforts in this area one can mention the SWWS 1 (Semantic Web and Web Services) project, which is intended to provide a comprehensive Web Service description framework, to define a Web Service discovery framework, and to provide a scalable Web Service mediation. 4 Agent-driven Semantic Web-based approaches to EAI According to a vision of one of the leading industrial companies [26], in future the infrastructure for management of company s will evolve to the general form shown on Figure 4: 1

9 Title Figure 4 Vision for an enterprise information infrastructure [26] As we can see, industrial companies seriously consider agent-oriented approach as a basis for integrating their distributed enterprise applications. The global condition-monitoring network is formed in a tree form. The global service center is the root and regional service centers are connected to it and industrial sites are connected to regional service centers. The industrial sites are all independent. Each of them has a connection only to the service center hierarchically next to it. Independence means also the possibility to detach from the tree at any moment, and work alone. However this disables also all the benefits of global condition monitoring. The global service center represented by Field Agent Server and industrial sites connected to it are shown (in a simplified form without regional service centers) in Figure 5 [27].

10 Author Figure 5 - Main components of a local Field Agent. [27] Field Agent is actually a collection of specialized software agents rather than one single agent. Local system architecture is thus modular based on agent technologies. For every task there is a specialized agent, which is at any time replaceable by a new version without disturbance to other agents. Another key advantage of the agent structure is its independence of the physical environment, making it easy to adopt the concept to different environments and needs. Agent environment enables also an easy way to add new features to the system simply by adding a suitable agent. Agents currently specified include interface agents ( HTML reporting, reporting and DDE reporting in Figure 5), information and task agents taking care of learning, analyzing and database

11 Title services ( Learning and Analyzing blocks in Figure 5), security agent and an agent taking care of field device monitoring. [27] Industry goes to a fact that in Plant Automation Systems of new generation human participation is minimal. Human users are responsible only for the acknowledgement of important decisions made by the machine. Thus most evidently Web Services of new generation will provide services for intelligent machines, not for humans. They will provide information support of the decisionmaking process performed by Smart Devices. However, how this information should be presented to become suitable for consumption by an intelligent Field Agent? It depends on the inference engine that the agent is based on. If the agent belongs to a type the inference of which is based on Knowledge Base, this agent will need from Web Service a knowledge that it is lacking for. According to FIPA 1 standards such information message can look like the one shown in Figure 6. The FIPA message is written in Agent Communication Language (ACL), such as FIPA ACL standardized by FIPA organization. The content of the message is expressed with specific content language, e.g. with Semantic Language (SL) that is standardized by FIPA. 1 The Foundation for Intelligent Physical Agents (FIPA)

12 Author Figure 6 - Example of message from Web Service to Field Agent One of the most important parameters in this message is the parameter ontology. It denotes the ontology(s) used to give a meaning to the symbols in the content expression [28]. The concept Ontology is extremely important in providing interoperability between intelligent entities. To make possible business between Web Service and Smart Device, the former must understand what it is asked for and the latter must understand the sense of information it is given. Industrial ontologies, standardizing terminology, can make a great contribution to the interoperability between intelligent systems of arbitrary operational scale: plant, enterprise, inter-enterprise. Since terms and their definitions are represented in ontologies in a natural, human-understandable form, it also makes easy the provision of interoperability between human (plant operators) and Field Agents. An approach that provides interoperability between agents by the creation of common problem domain ontology was applied by researchers in the development of the InfoSleuth system a significant component of the Environmental Data Exchange Network [29].

13 Title InfoSleuth has been designed as an agent-based system to integrate heterogeneous, distributed information sources and tools via the use of common ontologies. In other words, a set, or community, of InfoSleuth agents collaborate at a semantic level to execute information gathering and analysis tasks, where the underlying information sources can be diverse both in their structure and content. The agents communicate and reason about each other's capabilities in terms of a shared ontological model of information management to resolve user requests. Requests are posed in terms of an ontology, called the domain ontology of the application," that provides a semantic framework for information activities in the domain of the user's interest. The ontology used in the EDEN pilot project focuses principally on the relationships between contaminated sites, the wastes that cause the contamination, and technologies used to remediate specific kinds of contamination in specific media at each site. Other researchers have already made efforts of formalization of an industrial domain into an ontology, particularly for the oil-refinery domain [30]. After an intensive interview with domain experts, they found human operators use different terms to denote the same object depending on the context. The way of their word usage was carefully analyzed and it was discovered that it was not random except in a few cases. Many of the wordings had good justifications. The ontology approach allows existence of synonym names for the same notion. All the above facts lead to a possibility that if different industry enterprises separately engineer ontologies for the same problem domain, there can be inconsistencies between them. And when once they have to communicate, they will need a mechanism of inter-ontological relations establishment to enable interoperability. As a matter of fact for the same problem domain different ontologies can be created also for another reason. They can have a different depth of detail and represent different aspects of the problem domain. It depends on the tasks that will be performed by intelligent systems utilizing those ontologies.

14 Author What could an industrial plant ontology look like? The following section is an example of an ontology that was developed by Japanese researchers in [30] for an oil-refinery plant to standardize the communication vocabulary between plant agents. 5 Conclusions Industrial integration is closely related with the integration of industrial Web Services. Taking into account the future visions of industrial information infrastructure industrial Web Services have been recognized as services for Smart Devices (Field Agents). Therefore, industrial enterprise integration results in automated knowledge sharing between agents as representatives of different enterprises. For clear understanding of these knowledge exchange processes the structure of plant agent community has been studied. Such systems are selforganizing, intelligent, self-learning and fault-tolerant. Since the question of interoperability between industrial agents on knowledge level has a crucial value, the concept of ontology is discussed in this context. The evidences of beneficial utilization of ontologies in the interoperability issues have been given. The benefits they provide are: common vocabulary of industrial objects definitions, explicitation of expert knowledge, systematization, standardization and metalevel functionality. Some real-life projects where the potential of ontologies was fully utilized for provision of interoperability between information systems were described. The implementation of Semantic Web concepts to organize all the industrial information resources is expected to improve the resource sharing. To be suitable for automated machine search and other automated processing every resource has to be supplied with a machine-understandable description on a basis of a common ontology. Further analysis has shown that industrial information resources are highly heterogeneous and have very different types. The potential of Semantic Web will be fully utilized only when all the semantics (resource

15 Title descriptions, expert knowledge, whatever necessary) will become explicitly available. 6 References [1] Blueprint company, official web-site, last accessed 1st of November [2] R. Khanna: Top Challenges in Integration Projects, White Paper, ITtoolbox EAI Enterprise Application Integration section of ITtoolbox - online information provider on ICT, 21st May [3] Enterprise Application Integration: the Risks Behind the Rewards, White Paper, BMC Software, 11th December [4] A. Apte: Java Connector Architecture: Building Enterprise Adaptors, Publisher: Sams Publishing, ISBN: , p [5] J2EE Connector Architecture Specification, v. 1.5, Sun Microsystems Inc., 24th November [6] Duivestein S., Remmerde M. Web Services and the Insurance Business, July 2001, article on the Web, content/articles/marcosander01.asp. [7] Basex Inc., Executive report: Clerity for Enterprise Knowledge Sharing, January [8] Hohpe G., Enterprise Integration Patterns, ThoughtWorks Inc., [9] W3C Working Draft, Web Services Architecture, [10] W3C Recommendation, SOAP Version 1.2, Part 0: Primer, [11] W3C Working Draft, WSDL Version 1.2, [12] UDDI Spec Technical Committee Specification, UDDI Version 3.0, [13] Sliwa C. From e-speak to Web Services", Computerworld, December 2, [14] Sycara K. Autonomous Semantic Web Services. 15th Conference on Advanced Information Systems Engineering, Velden, Austria, [15] Web Services Flow Language, WSFL 1.0, Prof. Dr. Frank Leymann, Distinguished Engineer, Member IBM Academy of Technology, IBM Software Group, May 2001, /solutions/webservices/pdf/wsfl.pdf

16 Author [16] Sarginson M., Online instrumentation monitoring, impact on safety, uptime and process profitability, Metso Automation, Field Control Systems Committee, Exhibition [17] Pyötsiä J., Cederlöf H., Remote Wireless Presence in Field Device Management, ISA Proceedings [18] Thaden U., Siberski W., Nejdl W. A Semantic Web based Peer-to-Peer Service Registry Network, Technical Report, [19] Berners-Lee T., Hendler J., Lassila O. The semantic web. Scientific American, 284(5), 2001, pp [20] W3C Candidate Recommendation, Resource Description Framework (RDF), Schema Specification, 1.0, 27 March 2000, [21] Joint US/EU ad hoc Agent Markup Language Committee. Reference description of the DAML+OIL (March 2001) ontology markup language. March [22] The DAML Services Coalition, DAML-S: Semantic Markup for Web Services, May 2003, [23] Dean M., Connolly D., Harmelen F., Hendler J., Horrocks I., McGuinness D. L., Patel-Schneider P. F., and Stein L. A., OWL Web Ontology Language 1.0, Reference. July [24] Silicon Valley World Internet Center, Data-to-Cash (D2C): Business Opportunities Emerging from an Ocean of Data, Symposium, December 11, [25] Benjamins V. R., Contreras J., Corcho O., Gómez-Pérez A. (UPM). Six Challenges for the Semantic Web. ISWC2002, June, [26] Pyötsiä J. Future Care, ICT Development, Networked IT Solutions, Metso Corporation. [27] Ojala M. SW agent technology in global field device management, Helsinki University of Technology, Postgraduate seminar on automation technology, autumn [28] FIPA TC Communication. FIPA ACL Message Structure Specification, SC00061G, Geneva, Switzerland, March, [29] Fowler J., Perry B., Nodine M., and Bargmeyer B. Agent-based semantic interoperability in InfoSleuth, SIGMOD Record 28:1, March, 1999, pp [30] Mizoguchi R., Kozaki K., Sano T., Kitamura Y. Construction and Deployment of a Plant Ontology. EKAW 2000: