ASPASIA Knowledge Systems

- INFRAWEBS I -
Intelligent Framework for Networked Businesses and Governments Using
Semantic Web Services and Multi-Agent Systems

In this section the conceptualization and design of the ICT framework INFRAWEBS is introduced and illustrated - an intelligent framework for supporting organizational networking and the effective sharing of resources.
The challenges for this ICT framework are to:

These objectives are achieved by developing software components consisting of:

The framework mirrors the modern and future oriented way of ubiquitous computing, where computers will be relegated to the background of our lives and everybody may be at the same time client, broker and, provider of services.

RTD - State of the Art

Networked government initiatives are currently addressing many issues in citizen public sector interactions, for example in:

Administration processes become more open as business and citizens can find the information relating to their interests in network services. As a result, it is possible to create more active participation in decision making through the use of location and time independent network services. The administration processes become more efficient in terms of reduced manual work, which creates costs-savings in fixed operational costs and improves quality via automation. eGov portals are currently being built with the understanding that citizens and organizations can visit one stop services where they can find the information and services they need to fulfil either their obligations or pursue time and location independent services which earlier were only accessible at government or municipality offices.

Further, portals have adopted metadata technologies in assisting users to find the information easier than using fulltext search. In fact, metadata give them the ability to efficiently find the relevant information from a mass of otherwise unstructured information. The accessibility of services is, however, only starting to take advantage of the latest industry standards such as SOAP, WSDL, and UDDI. These three technologies are often associated with "Web Services" concept, where services implemented with various technologies can be interfaced with other systems using open standards independently of programming languages and architectures. The full power of service integration into e-business services can be unleashed by combining the best of semantic descriptions i.e. metadata and the technical interfacing of services i.e. the semantic web services technologies.

Networked businesses have only recently been able to benefit from open standards based integration between various architectures. The Web Services family of technologies is often associated to bridging Microsoft and Java J2EE platforms but it has, in fact, more far reaching scope. The Web Services stack of technologies has been widely implemented from the SOAP, WSDL, and UDDI conformance point of view. However, the networking of more than two end points (one client and one server), has only been suggested by IT vendors in specifications without actual implementations. BPEL-4WS is an example of a business process specification that addresses the participation of multiple services into a single client request. In this setting, the transactionality and quality of service (QoS) issues become critical as services need to be run with a predictable end state and they need to be available for business critical execution.

It could and should be determined that there is lack of semantic based systems using decision supported techniques - tools and systems, which enable an "easy to use" web service designing and composing interface. At present the establishment of semantic based software components is reserved to specialist and programming experts. Furthermore, systems are missing for effective and time optimal collaboration in reconfigurable networks for installing and maintaining business and governments" alliances and partnerships. The present software tools and techniques are mostly incoherent - due to proprietary aspects and features that prevent an effective adjustment and balancing of information and knowledge pools. Ontologies provide, in this context, some kind of "normalization" of knowledge spaces - a base precondition for successful communication and collaboration in business and governmental areas.

The market of network business software is mainly characterized by existing different types of environments using open exchange and communication standards, like XML, RDF, RDF(S). Coupling and communication between these proprietary business solutions is still retarded and impeded due to missing adjusted and standardized formalization resp. normalization features for comprehensive knowledge pools. It seems that the approaches of the Semantic Web (SW) especially the normalizing character of the ontology approach should be the solution. This also seems reasonable, because in the scientific research and development field some kind of rush is to be detected and to be observed: since the activities of the DAML oriented international working groups a massively growth of interests is to be noticed.

There exists already network collaboration software components in e-business and e-government. But the main lack is that they don"t "understand" each other in a proper and effective way. The semantic especially ontology based development should be the right solution to overcome these communication and understanding gabs and lacks.

INFRAWEBS Methodology

The framework is aimed to develop a software platform where business processes will be run according to open standards such as BPEL4WS once it has been ratified by a relevant standards body. In addition, it is aimed to develop software components to automatically determine which business partners are needed to be invoked, in order to run a complete business process. To this end, a bridging of Web Services and Semantic Web technologies has to be implemented in the context of rich semantic service descriptions. Such descriptions are useful for service developers and for companies who wish to "plug-in" their IT systems to automatically run networked business between known or previously unknown business partners. The INFRAWEBS approach provides intelligent tools and systems for optimizing and enhancing dynamic collaboration and networking.

This ICT framework supports different organizations in their organizational networking, process integration, and sharing of resources. It is designed as a flexible and reconfigurable platform, which enables the organizations to build faster and more effective partnerships and alliances. It is one further step in the direction of implementing the vision of virtual collaborating entities - in sense of "fast reaction and interchange in virtual enterprises and governments". Virtual enterprises could be established which consist of several building blocks, modules, products, and services which are interchanged, modified and optimized within partnerships and alliances. Being directly connected to business partners and having direct access to their services and products, advances the stimulation of innovation through positive effects on the knowledge value chain.

The framework increases the human effectiveness and capacity for reinforcing competitiveness. Huge costs in relation to human working hours and resources are obvious. Agent based semantic web services will reduce information stray and redundancy and will have a positive effect on users and customers. The application of semantic web tools will improve the commercial use of the web and enhance the level of confidence between negotiation partners. This will broaden the use of the internet as a self evident tool of commerce and will reduce market inefficiencies.

One such inefficiency arises through "information asymmetries" - by the implicit promise of good information serving with the best value. Price dispersions are found for equal products requests (contractors searching for the best value at a present market situation). As a matter of fact, "the more research invested in finding high and low priced items, the larger the price scale". Naturally, as "the amount of data increases, the effort and time of the retrieval also increases".

From the technical point of view, one of the important innovation related aspects, with respect to the INFRAWEBS platform, is the integration of semantic based structures and services in networking and collaboration. The integration of such semantic related structures increases the communication capability of different heterogeneous systems and platforms. Using these methodologies and techniques makes the systems "understandable" to each other (provided for interoperability) - not only in sense of applying standards and formalizations, but also automated and agentbased knowledge acquisition and collaboration exchanges will be provided and enhanced.

Due to the implementation of "intelligence" within the decision making and supporting facilities, it serves in automating the processing of semantic structured knowledge and assists in building new products and value added services along the value chain. Due to this fact, no specialized developers or personnel are necessary or to be involved in service and product configuration and composition. Furthermore, the integration of tools for handling and considering "experience" and "learning from the past" (Case Based Reasoning - CBR) optimizes the composing and generation of services.

In this sense, the INFRAWEBS ensures the coevolution of technology and applications. Through this new basic entity, specific services will be exploitable in combination with capabilities to incorporate services provided by the business or governmental partners. Therefore, a precondition is given in establishing innovative products and services.

The reduced scheme of the INFRAWEBS is illustrated in Fig.1. The framework consists of two main components: a knowledge brokering component and a client & provider component, according to the formulated vision in the previous section.

Fig. 1: The reduced scheme of the INFRAWEBS consisting of a broker and client & provider part

The knowledge broker is realized as an organizational memory (OM) with decision support facilities. This broker is designated to endow the specific entity (business enterprise or governmental department) with comprehensive facilities to maintain and process the entity specific knowledge (knowledge artefacts). This knowledge management component is enhanced with decision support tools, especially in project handling and management.

The client & provider component is designed as a semantic and ontology based module handling and maintaining semantic web services - a Semantic Web Service Unit (SWU) as a collaboration and knowledge sharing component. It acts in two directions:

It provides web services to the ordinary user and customer of the entity. The use of this web services by the customer constitutes and supports the daily business and governmental transactions. The main B2C and G2C process is conducted via this channel. It enables via SW services a comprehensive exchange and incorporation platform among business or governmental partners and provides the network with interoperability properties.

Fig. 2: Implemented in different entities, either the business or the government area the INFRAWEBS approach enables modern types of networking and dynamical collaboration

As illustrated in Fig.2, the networking partners provide SW services, which are incorporated in their own services palette (product range). On the one hand, it is an expansion of the own services, on the other hand, the creation of new products and services is facilitated to develop value added products and services. The partners are endowed with facilities to optimize their partnership relations; furthermore, they are enabled to build faster and more effective partnerships and alliances with new business or governmental partners. This reflects an innovative and modern type of semantic based networking (semantic based network architecture).

The use of semantic based structures and ontologies [17] guarantees a formalized normalization of different knowledge domains and their automated processing. The usefulness of the framework is in the ability of the user to be "broker, provider and client" at the same time - within one system. There is no need to use different software components or packages - the user is embedded in an environment, which allows to change the action type and modus according to the given demands. This framework enables the user to act and react in a "multimodal" manner (Fig. 2).

INFRAWEBS Architecture & Modules

The development goal of the system is the design and realization of a framework that fulfills, among others, two essential tasks:

a) decision supported brokering of entity specific know-ledge (broker), accessed either via partner networks or in interchange with customers and clients;

b) enabling dynamic collaboration and semantic based knowledge and service sharing within partner networks and decision supported service advertising in customer oriented areas (client & provider).

INFRAWEBS Architecture

As illustrated in Fig. 3, the system architecture of the INFRAWEBS is designed using a two component ap-proach - a knowledge brokering component (OM/SIR) and a SW Service Unit (SWU) - divided into several modules. The broker component consists of the organizational memory of the entity (business enterprise or govern-mental department) coupled with a decision supported project management module. A Semantic Information Router (SIR) interconnects the broker component and the client & provider component by accessing the se-mantic web repository module of the SWU. Coupled with this repository are three modules for

The designing, composing and execution modules are sup-plemented by a multi-agent module. Security and pri-vacy aspects are considered in a further module con-nected to the execution and agent related modules.

Fig. 3: System architecture of the INFRAWEBS - designed as a two-component framework

INFRAWEBS Modules

OM/PM - organizational memory and decision supported project management

The OM/PM module is realized as an organizational memory (OM) with a supplemented decision supported project management tool (PM). The OM approach de-livers tools that integrate the decentralized resources (either across the web or the alliances network re-sources) into a coherent corpus of interrelated informa-tion: the entity specific knowledge artefacts are proc-essed and managed according to the entity specific users demands. Accessing the knowledge corpus of the OM the PM extracts relevant knowledge objects for proc-essing within project management tasks. The design of the PM is oriented on the TECB (task extraction and correspondence browser, KNIXMAS project [18]), en-hanced according to semantic and ontology structures. CBR methodology is used in the decision making ap-proach - to consider experience aspects, in the sense of the con-ceptual approach of experience factories.

SIR - semantic information routing

The SIR module extracts the relevant knowledge arte-facts out of the OM and processes a conversion proce-dure to transform and adapt the provided knowledge objects to an ontology specific structure using RDF / XML and further meta-data description schemes. The knowledge objects are semantically preprocessed for delivery to the SW repository. Furthermore the SIR is enhanced to access different types of OMs, due to the fact that in several business and governmental organiza-tions different legacy systems exist. First a reduced model of the SIR is realized by handling simple knowl-edge artefacts such as pure text objects. After that a gen-eralization is performed to consider different types of OMs (Fig.4).

SWS-D / SWS-C / SWS-E - semantic web service designer, composer and executor

These three modules access the SWS-R (repository) for single SW service descriptions. The designer and the composer are realized as CBR based SW services devel-opment tools and are endowed with decision-making capabilities (Fig.5). The SWS-Designer processes a designing procedure for single static web services ad-vertised to the external (customer) as well as in the in-ternal (network-ing) area. The SWS-C - the composer - is responsible for the dynamic composing of available (existing) SW services. This SW services might be lo-cated and de-signed within the actual INFRAWEBS or loaded and accessed via the collaboration network and incorporated from other SWUs - a composing process by combining existing "own" and "foreign" SW serv-ices, provided by the collaboration partners (partner INFRAWEBS). The SUA is inquired to process agent based discovery and mediation actions to provide ade-quate SW services for composition purposes. The de-signing and composing process is throughout decision supported (CBR - case based reasoning approach), es-pecially under the aspect of "learning from the past". Accordingly, a self-organ-izing case base is imple-mented for storage and retrieval of designing / compos-ing expe-rience patterns. The SWS-E - the SW service executor - maintains and supervises the SW service exe-cution on the network side as well on the customer side.

SUA - service and user interface agent environment

This module is responsible for both action sides - the relation to the customer, the external user of the web service and the internal - the networking collaboration area. On the internal - the networking and collaboration side - it provides facilities for discovery of SW services. Furthermore it is called in case of inconsistencies related to the composing aspect of SWS mediation facilities (Fig. 5). This mediation is necessary because in "real life" scenarios, the situation where two agents share one ontology or vocabulary in its entirety is the exception and not the rule. Therefore, this mediation facility will mainly consider situations, where agents share at least a part of their vocabulary and a mapping of the other con-cepts or vocabularies has to be arranged. In this context the agent interaction with the web services are real-ized through OAA (Open Agent Architecture). On the user side it provides a retrieval and execution interface for advertised SW services. The user/customer mandates a user interface agent for fulfilling the service demand. The user/interface agent gives recommendations based on the user"s query. Considering the given user service de-mands, the interface agent triggers service agents for discovery and selection of advertised web services.

SP - security & privacy

The security/privacy module represents symbolically the implementation of several security and privacy preserv-ing techniques and specifications. Beyond the standard security aspects, such as firewalls, authentica-tion and encoding, the specific security issues related to the upper layer of semantic web service structures has to be considered. This includes the important issues related to preserving privacy. A privacy model is established that is structured in user, service and data privacy. Ac-cording to the developed model, specifications and add-ons are processed and applied in the relevant mod-ules, especially in the SW service execution and the agent module. The development and implementations are oriented on the directives and guidelines pro-vided by the Web Services Security TC (WSS TC) of the OASIS - Committees, especially the detailed de-scription of the OGSA security architecture given in the "Open Grid Services Security Architecture" document. (security roadmap - OGSA Security Arch.) together with the Roadmap (white paper) from IBM Corporation and Microsoft Corporation.

Fig.5: The interrelation of modules in the SWU: designing, composing and execution using a self organizing case base ("learning from the past")

The INFRAWEBS approach provides intelli-gent tools and systems for optimizing and enhancing dynamic collaboration and networking. The application of semantic web services in conjunction with knowledge representing ontologies ensures the feature of reconfigu-rability and interoperability in a modern effective way. The overall methodology used to reach the development goals is described. To guarantee a user comfortable handling of this collaboration structure, several supporting tools have to be realized to assist the "non-expert" user in execution of his problem and sharing relevant working actions, such as SW (semantic web) services designers, dynamic composers and executors. A multi-agent-system pro-vides on the one hand user interface agents on the other hand agents to perform SW service discovery and me-diation.

[1] T. Berners-Lee, J. Hendler, and O. Lassila. "The Semantic Web" Scientific American, 2001.

[2] D. Fensel, "The Semantic Web and its Languages", IEEE Intelligent Systems, Trends and Controversies, pp.1, November/December, 2000.

[3] O. Lassila and R. Swick, "Resource Description Framework (RDF) Model and Syntax Specification", W3C Recommendation, World Wide Web Consortium, Cambridge (MA), February 1999; available on-line as http://www.w3.org/TR/REC-rdf-syntax/

[4] S. Mcllraith, T.C. Son, H. Zeng, "Mobilizing the Semantic Web with DAML-Enabled Web Services", Proc. of the 2nd Int. Workshop on the Semantic Web - SemWeb"2001, pp.82-87, May 1, 2001, Hongkong, China.

[5] J. Hendler, "Agents on the Web", IEEE Intelligent Systems, Nov/Dec 2001

[6] D. L. Martin, A. J. Cheyer, and D. B. Moran, "The open agent architecture: A framework for building distributed software systems", Applied Artificial Intelligence, vol. 13, pp. 91-128, January-March 1999.

[7] S. McIlraith, "Modeling and Programming Devices and Web Agents", Proceedings of the NASA Goddard Workshop on Formal Approaches to Agent-Based Systems, Lecture Notes in Computer Science, Springer-Verlag, 2001.

[8] K. Sycara, M. Klusch, S. Widoff and J. Lu, "Dynamic Service Matchmaking Among Agents in Open Information Environments", Journal ACM SIGMOD Record (Special Issue on Semantic Interoperability in Global Information Systems), Vol. 28, No. 1, pp. 47-53, March 1999.

[9] Andreas Eberhart, "An Agent Infrastructure based on Semantic Web Standards", Elsevier Science, 2002.

[10] K.M.Gupta, D.W. Aha, & N.Sandhu, "Exploiting taxonomic and causal relations in conversational case retrieval", Proceedings of the Sixth European Conference on Case-Based Reasoning pp.133-147, Aberdeen, Scotland, Springer, 2002.

[11] R. Weber and R. Kaplan, "Knowledge-based knowledge management", in: Faucher C., Jain L., Ichalkaranje N. (eds), Innovations In Knowledge Engineering, Physica-Verlag, 2003.

[12] R. Waldinger, "Deductive Composition of Web Software Agents", Proceedings of the NASA Goddard Workshop on Formal Approaches to Agent-Based Systems, Lecture Notes in Computer Science, Springer-Verlag, 2001.

[13] Srini Narayanan, Sheila A. McIlraith, "Simulation, Verification and Automated Composition of Web Services", WWW 2002, May 7-11, 2002, Honolulu, Hawaii, USA.

[14] J.Segovia, P.S.Szczepaniak, Niedzwiedzinski, M. "E-Commerce and Intelligent Methods", Springer, 2002.

[15] Marling, C., Sqalli, M., Rissland, E., Muñoz-Avila, H., & Aha, D. "Case-based reasoning integrations", AI Magazine, 23(1), pp.69-86, 2002.

[16] I.Watson, & D.Gardingen, "A Distributed Case-Based Reasoning Application for Engineering Sales Support", Proc. 16th Int. Joint Conf. on Artificial Intelligence (IJCAI-99), Vol. 1: pp. 600-605, Morgan Kaufmann Publishers Inc. 1999.

[17] F. van Harmelen and I. Horrocks, "FAQ"s on OIL: the Ontology Inference Layer", IEEE Intelligent Systems, Trends and Controversies, pp. 3-6, November/December 2000.

[18] H.-J. Nern et al, Final Report: "Knowledge Shared XPS Based Research Network Using Multi Agent Systems", EU-FP4, Esprit/Inco 977113