Modeling Web Services Policy with Corporate Knowledge
Dong Huang
Siemens AG, Corporate Technology Otto-Hahn-Ring 6
81739 Munich, Germany [email protected]
Yi Yang, Jacques Calmet
Institute for Algorithms and Cognitive Systems University of Karlsruhe (TH)
76131 Karlsruhe, Germany {yiyang,calmet}@ira.uka.de
Abstract
Web Services Policy is used to specify service con- straints. Although many efforts like WS-Policy have been done to provide declarative configuration languages, it is still hard to discover and exchange the configuration infor- mation in the service computing environment. A novel ap- proach to share service knowledge and application-specific information is needed. In this paper, we model web ser- vice policy with corporate knowledge, which is defined as the amount of knowledge provided by individual agents.
This approach provides a distributed knowledge manage- ment method to our proposed semantic policy framework that enables reasoning, which is necessary for policy cre- ation, conflicts resolution and negotiation.
1. Introduction
In recent years, a significant number of companies have implemented e-business solutions because an investment in e-business technologies provides the promise of a com- petitive advantage through lower transaction costs and the integration of processes. Companies are enclosing tradi- tional computing tasks, such as database access or com- mercial transaction systems, in wrappers as software ser- vices to connect them to the Internet. New tasks, such as computerized auctions and e-marketplaces are also intro- duced as business services. The concept of Web Services is thought to be the next generation of e-business architectures for the web. Such a Service-Oriented Architecture (SOA) describes principles for creating dynamic, loosely coupled systems based on services, but no single specific implemen- tation.
Both Web services and Semantic Web services consider services as fundamental, central entities. Service descrip- tions play a central role in publishing services for use, cat- egorizing them and finding the right service to satisfy a need. In [26], three different aspects are considered when
talking about service descriptions: (1) functional, (2) be- havioral and (3) non-functional. The functional description contains the formal specification of what exactly the ser- vice can do. The behavioral description contains the for- mal specification of how the functionality of the service can be achieved. The non-functional description captures con- straints over the previous two. The most important issues of the non-functional descriptions are security, authentication, and privacy issues related to the exchange of information necessary for the service to be consummated and legal or contractual issues between the provider and requester of the service.
A policy is information which can be used to modify the behavior of a system [18]. Policies describe the capabilities, requirements and constraints of a service. Non-functional service descriptions are often used to formalize or specify assertions in policies.
Two problems arise from the application of web services in e-business:
1. Change management. Nowadays e-business scenarios always involve more than one organization. Changes to business rules should take effect in an efficient way without recoding or stopping the system.
2. Gaps between Business and IT.Managers and admin- istrators, who work at the business level, tend to use abstract, high-level and human-readable policies to specify requirements or describe the services. At the IT level, one works the other way around; policies are usually specified in a machine-readable way and en- forced according to the specific environment. Knowl- edge exchange and sharing is necessary in this case to build an intelligent service.
To solve these problems, a security policy framework for business processes was proposed in [15], which gave a preliminary idea of policy management and implementa- tion in SOA. In a new line of work, this framework is fur- ther extended and a semantic policy framework for service-
oriented computing (SOC) is introduced. The main contri- butions of the new framework are:
• It allows knowledge exchange and sharing among all stakeholders in e-business scenarios, which are defined as service providers (P-agent), service requester (R- agent) and middle-agents (M-agent) [8].
• It provides a novel way to represent policies and to rea- son over them.
This paper focuses on the first contribution. Agent Ori- ented Abstraction (AOA) [5] addresses knowledge manage- ment issues by viewing an e-business services-oriented net- work as a multi-agent paradigm for distributed computa- tion. The concept of a virtual knowledge community (VKC) enables us to model corporate knowledge as the amount of knowledge provided by individual agents [19]. In our ap- proach, corporate knowledge can provide the basic informa- tion model for policy specification and conflict resolution by capturing business rules and the non-functional require- ments of web services.
The remaining sections of this paper are structured as follows. Section 2 scratches the surface of the agent ap- proach for web services and introduces the principle of AOA and VKC. Section 3 gives an overview of our seman- tic policy framework for web services. Section 4 introduces an e-business scenario and shows some details on how to model policies based on corporate knowledge. Section 5 surveys related works. Section 6 outlines some future re- search and concludes.
2. Agents, AOA and VKC
This section outlines several different, although inter- related, agent-based viewpoints for web services, agent ab- straction and corporate knowledge.
2.1. Agents and Web Services
In the cyber world, agent-based approaches are more powerful when running agents in a distributed and dynamic environment (potentially on a web-wide scale) to perform complex actions for their users [12]. Uniting agents and web services can enhance the construction and flexibility of web service applications [27].
Before introducing other agent-based approaches in fol- lowing sections, it is necessary to remind some agent related definitions first. An agent is capable of autonomous action in assigned environment in order to meet its design objec- tives [30]. Based on this definition an intelligent agent can be extended with three additional characteristics: reactiv- ity, proactivity and social ability. The concept of multiagent has emerged as a paradigm for designing complex software
systems. It is mainly used to better formalize problems in Distributed Artificial Intelligence (DAI) [29].
The world of web services is characterized as loosely- coupled distributed systems based on SOC. The use of web services could be considered as actions that the agent might execute to meet its goals. In [17] four major trends in in- ternet computing were analyzed which have driven SOC and Multi Agent Systems (MAS) research into the future.
They are among emerging approaches with MAS-like char- acteristics in SOC, such as ubiquitous computing, ontolo- gies, service-level agreements and quality-of-service mea- sures for instance. All of them can be suitably tackled with MAS concepts and techniques.
2.2. Agent Oriented Abstraction
The agent oriented abstraction paradigm is introduced in [5]. It is a high level abstraction for agent modeling and covers the concepts of agents, annotated knowledge, util- ity functions and society of agents. Indeed, AOA relies on Weber’s classical theory in Sociology [28].
In the approach of AOA, agents are seen as objects and through their knowledge contents, they are organized into annotations that gather classes. Encapsulation, inheritance and polymorphism are features that can be adequately de- fined. The AOA model can be abstractly summarized by a number of basic definitions. A detailed instruction of that will not be given here, but is to be found in [5].
Chiefly, in AOA all agents have two parts: a decision mechanism and knowledge. For the former, a scope of possible classification of utility was given: expected util- ity function, the common sense measure of usefulness, the class of models and the class arising from logical modeling.
For the latter, there are also some associated classes: on- tology, communication, cognition and safety. Based on the knowledge annotations, agents can generate utility related to its tasks and goals.
Within the AOA approach web services and their related policies can be abstractly modeled in the knowledge part of agents running around the semantic web. Section 4.2 will illustrate this approach with more details through a working example.
2.3. Virtual Knowledge Communities In [19] the application of the AOA model to the abstract modeling of corporate knowledge is investigated. To avoid the separation between agents and knowledge, it was con- sidered that agents have explicitly represented knowledge and communication ability.
Traditionally, information is mostly centralized within a uniform information structure. This view point is not truly compliant with the nature of knowledge that is subjective,
distributed and contextual [3]. From the perspective of the knowledge information society, modern knowledge man- agement often focuses on the constitution of communities of practice and communities of interest [10].
The concept of a community of practice or a commu- nity of interest can be supported in a virtual community in order to bring the concerned agents together to share their knowledge with each other. A community is a place where agents can meet and share knowledge with other agents which share a similar domain of interest. The concept of a VKC was introduced as a means for agents to share knowl- edge about a topic [20]. It aims to increase the efficiency with which information is made available throughout the society of agents.
From the point of view of corporate knowledge manage- ment, agents can be individuals, software assistants or au- tomata. Agents possess knowledge and processes within the society tend to make agents produce and exchange knowledge with each other. These processes are distributed throughout the society and contribute through their own in- trinsic goals to solve a unique high-level challenge. This provides the link between corporate knowledge and VKC [19].
In the implemented model [11] there are two main mod- eling types for VKC: agent modeling and community mod- eling. The former has four key notions: personal ontology, knowledge instances, knowledge cluster and mapper. Per- sonal ontology represents the knowledge of an agent. It de- scribes the taxonomy of the relationships between the con- cepts and predicates that an agent understands. The knowl- edge instances are instances of objects defined into the per- sonal ontology. It was assumed that an agent’s knowledge consists of both its personal ontology and knowledge in- stances according to its personal ontology. The knowledge cluster is that sub-part of an ontology that can be shared among agents. Clusters are defined by their head concept, a pointer to the different parts of knowledge existing in a cluster. The mapper chiefly contains a set of mapping from personal terms to mapped terms, and allows an agent to add such mappings, and use the mapper to normalize or per- sonalize a given knowledge cluster or instance using these mappings. It facilitates knowledge sharing among agents with regards to the heterogeneity of knowledge.
Community modeling also has some key notions: do- main of interest, community pack, community buffer. A do- main of interest exists in each VKC and is similar to the con- cept of ontology for an agent. It is given by the community leader which created the community. The community pack is what defines the community. It consists of a community knowledge cluster, a normalized ontology which contains at least the head of the community cluster, and the identi- fication of the leaders of the community. The community buffer can record messages which are used by the member
of a community to share their knowledge. This approach is compatible with blackboard systems, but still has its dif- ference, because agents cooperate to solve their respective problems, not for a unique goal.
The VKC approach has been designed and partially im- plemented as a prototype system. The implementation is based on Java Agent Development Framework (JADE) and Java Runtime Environment (JRE) platform. It was tested and evaluated. A component of the system enables us to simulate virtual knowledge communities (VKCs).
3. Semantic Policy Framework
In this section, a semantic policy framework is intro- duced for the specification and enforcement of policies to support security and trust management for SOC.
3.1. Security Policy Specification Language Various approaches have been done to achieve secu- rity policy specifications, including logic-based languages, role-based access control, various access controls and trust specification techniques [7]. But, a specification language, which can meet the following requirements, is still missing.
• Support of non-functional service descriptions.
How to model non-functional properties into the poli- cies and enable reasoning over them?
• Integration of Business Rules. Business rules state core business policies. They control and influence business behavior. How to integrate them with the knowledge base and specify policy using rules?
Various web services and semantic web services approaches such as UDDI1, OWL-S [6], SWSF [2] and WSMO/WSML [9] have been investigated to describe the non-functional properties of a service. In [23] a set of the most relevant non-functional properties for Web services and their mod- eling are described. An overview of all these approaches is given in [26].
Business Rules are used for categorizing facts important to a business. They also require or prohibit actions by a business. OMG2 has been widening its scope to include business modeling. Several of its recent requests for pro- posals have been about or related to business rules. These proposals are Semantics of Business Vocabulary and Busi- ness Rules, Production Rule Representation, Business Rule Management.
We define the policy specification language with an up- per ontology in [14]. From the upper ontology, a domain- specific policy ontology describes the vocabularies for poli- cies, business rule terms and service descriptions used
1http://www.uddi.org
2http://www.omg.org
Reasoner Service Repository
Policy Repository
Rule Repository
VKC
Semantic Policy Framework
Policy Management Tool
Policy Service
Knowledge Service
Knowledge Management Tool Response
Request
Figure 1. Framework architecture
within the domain. By using the domain-specific policy template, a policy specification can be generated automat- ically with Ontology language OWL [21] and Rules lan- guage SWRL [13]. In [14], the details about the specifica- tion and management of policies in this policy framework is introduced. In section 4, we will show how the knowledge sharing and exchange help us to model policies for services.
3.2. Framework Architecture
In order to support conflict resolution and life-cycle man- agement of policies, as well as enable reasoning over the knowledge base, an architecture of the semantic policy framework is illustrated in Figure 1. It includes two sup- porting services: a policy service and a knowledge ser- vice. Additional components include: reasoner, manage- ment tools and repository.
• Policy Service. The policy service acts as a policy de- cision point3(PDP) for web services policies, includ- ing security and QoS requests. The policy service acts on service requests and renders a decision.
• Knowledge Service. The knowledge service manages the VKCs and the repository of static knowledge base:
business rules, policies and service descriptions.
• Repositories. Three repositories are used to store ser- vice descriptions, policies and business rules. This
3http://www.ietf.org/
information can be managed through the policy and knowledge service.
• Reasoner. Reasoner is used to perform logical infer- ence over corporate knowledge based on the knowl- edge repositories and VKCs. In our implementation, we use KAON24 as reasoner. The major advantage of KAON2 is that it is a very efficient reasoner when it comes to reasoning with Description Logics ontologies containing very large ABoxes and small TBoxes [22].
The terms Abox and Tbox are used to describe two different types of statements in ontologies. Together Abox and Tbox statements make up a knowledge base.
• Management Tools. The policy management tool acts as the interface to policy service; it manages the life- cycle of policies, creates and deploys new policies.
The knowledge management tool provides an interface to manage the VKCs and knowledge repositories.
VKCs build a special knowledge component outside the framework and are managed by the Knowledge Service Agent, which also acts as a leader agent on the knowledge service.
As the policy service acts as a PDP, it receives the service request from the policy enforcement point (PEP) at the ser- vice end. There are two possible options: a) It finds out the suitable policies from the policy repository, renders the de- cision and sends the result back to PEP, b) It cannot find the proper policy or there are conflicts in policies, the request will be converted to a request to the knowledge service. The knowledge service will analyze the request and prepare a knowledge base for the next reasoning step. The knowledge base is built both on static knowledge from repositories and dynamic knowledge from VKCs. Based on the result of rea- soning on the knowledge base, the policy service can make its decision of the service request.
4. Modeling Corporate Knowledge
This section illustrates an e-business scenario, in order to show how to model policies based on corporate knowledge.
4.1. Scenario
To demonstrate the need for modeling policies within the Agent Oriented Abstraction, a scenario of the payment problem in the e-money system is as follows.
A mobile network operator has developed a service plat- form to enable the customer to pay bills and buy things from both online- and offline shops with their PDA, cell phone or laptop. The other customers of the service platform, for ex- ample a bank, will accept the payment request and decide
4http://kaon2.semanticweb.org/
if the transaction should be processed. Because of the dy- namically changing business rules, legal framework for e- business and customer preferences, policies for these issues are distributed and managed by different stakeholders in the transaction.
• A customer, Bob, buys a video camera in shop A at the airport B, issues a payment request via his PDA to a payment service of his bank on the service platform, specifying payment details including the amount to be paid and the payment method. For example, the pay- ment amount is $1,000 and the payment method is by Visa card.
• The payment service of the bank gets the transaction request. First, the bank will handle the request. At that airport only a transaction in a foreign currency like US dollars is allowed and the requestor has a good credit history. Second, based on the preferred payment method and physical connection type of the customer’s PDA, which could be WLAN, GPRS or Ethernet, an encryption algorithm will be chosen to protect the in- tegrity of the transferred data, because physical con- nection types can support different bandwidths. The bank has an agreement with the shop: request from customers in the shop at the airport will be handled with a higher priority. The priority ranking of policies is legality, security and then convenience.
• The network operator knows that the connection type of Bob’s PDA is GPRS and it is connected from a base station at the airport B.
• The local government allows trading in US dollars and the bank has checked Bob’s credit history to be good.
In this scenario, we have one authorization policy of pay- ment service, one obligation policy of PDA, three rules and one non-functional description of payment service. The concepts of authorization policy and obligation policy are introduced in [25]. We model three policies of the payment service: Legality Policy (Access control), Security Policy (Data Integrity) and Convenience Policy (QoS). These kinds of services will be different domains of interest in VKCs.
4.2. Agents and Their knowledge
In this scenario, there are five agents categorized as fol- lows: P-agent: “Bank”; R-agent: “PDA”; M-agents: “Net- work Operator”, “Local Government” and “Airport”. It is assumed that any of these agents is modeled as a VKC. This implies that an agent’s knowledge is ontologically struc- tured. Figure 2 gives a simplified view of knowledge of agents in the scenario. The dashed line indicates instances of concept, while the solid line with label and arrow indi- cates the relationship between concepts.
Knowledge of Network Operator Agent Knowledge of Bank Agent
Knowledge of Airport Agent have Shop Knowledge of PDA Agent
Encryption dependOn use
Bank have CreditRecord
{GOOD, BAD}
Services
accept Payment Service
VISA Credit Card Payment Method
Master Customer
employ Bob’s credit history
is provide
Shop A
GPRS WLAN
Network Operator
ConnectionType provide
Customer
Bob Alice have signedIn
Base Station
Airport B Encryption provide
DES RSA SAFERK64 dependOn
Shop A Bob
Owner PDA
Location hasProperties
PDA p
Government
Currency
Local
US Dollar
Airport allow
Knowledge of Local Government Agent
hasProperties
Figure 2. Agent’s knowledge with their per- sonal ontology and instances.
4.3. The VKC approach
The basic concepts of VKC were introduced in Section 2.3. Now we can illustrate in Figure 3 how to build corpo- rate knowledge with VKC based on the individual knowl- edge base shown in Figure 2.
Figure 3 depicts a simplified community of communi- ties for the e-business scenario consisting of possibly sev- eral member communities. Agents can create as many com- munities as they want. In this picture there are three virtual knowledge communities for different domain of interests:
legality, security and convenience. As mentioned above an agent can join more than one community, agents ”PDA” and
”Bank” participate in each of these three communities, re- spectively as service requester and service provider. The M-agents ”Local Government”, ”Network Operator” and
”Airport” joins the communities “Legality”, “Security” and
“Convenience” respectively.
In order to explain how a community generally works to facilitate knowledge sharing, the community “Security” is
Bank
Network Operator PDA
PDA Bank
Local Government
PDA Bank
Community of Communities
Convenience
Airport
Community Agent
Legality Security
Figure 3. The simplified view of virtual knowl- edge communities of the e-business sce- nario.
Bank
Customer:{Bob}
Network Operator Customer:{Bob}
ConnectionType:{GPRS,WLAN}
PDA
employ
accept
depentsOn
signIn
dependsOn CummunityBuffer Security
Owner:{Bob}
Encryption:{RSA}
Encryption:{RSA}
Service:{PaymentService}
PaymentMethod:{Visa,Master}
BaseStation:{Airport A}
Figure 4. Agents who shared their knowledge within the community “Security”.
PaymentMethod:{Visa,Master}
Customer:{Bob}
ConnectionType:{GPRS,WLAN}
employ
accept
depentsOn signIn
dependsOn Encryption:{RSA}
BaseStation:{Airport A}
Service:{PaymentService}
Figure 5. Knowledge evolvement of the agent
“PDA” following the knowledge exchange in the community “Security”.
illustrated in Figure 4 in detail. Agents “Bank”, “Network Operator” and “PDA” can read and write messages in the community buffer, in order to share their knowledge clus- ters and instances. The dashed line with direction from an agent to the community butter implies that an agent writes messages into the buffer, while the dashed line with in- versed direction implies reading messages.
In Figure 4 messages are structured through simpli- fied notations of ontologies with concepts, their instances and relations between them. For example, the notation PaymentMethod:Visa,Masterdenotes that the con- cept PaymentMethod has two instances: Visa and Master.
The agent “Bank” writes one message into the buffer and reads two messages which are coming from “Network Op- erator” and from “PDA” respectively. The agent “Network Operator” writes two messages and just has interest in the message from “PDA”. “PDA” inputs only one simple mes- sage but benefits from the contribution of other agents in this community. After the knowledge exchange step, each agent can update its previous knowledge base respectively to its knowledge cluster and personal ontology with the new information. The feature introduced in the above example (especially for agent “PDA”) is illustrated in Figure 5. Now
“PDA” knows that it can sign in the base station in “Airport B” and employs the “Payment Service” with Visa or Master card.
4.4. The Knowledge Integration
The aim of the knowledge integration is to build cor- porate knowledge with VKCs and knowledge from repos- itories defined in the semantic policy framework. Because
OWL is closely related to Description Logic, it has many features that come from the family of knowledge repre- sentation system [24]. So we adopt OWL to represent the knowledge in our approach. Although OWL is supposed to be effectively implementable, its expressive power is lim- ited. To solve this problem, SWRL extends the set of OWL axioms. The following code represents a simplified version of the PDA Agent’s knowledge after the Knowledge Inte- gration.
<owl:Class rdf:ID="Encryption" />
<owl:Class rdf:ID="PDA" />
<owl:Class rdf:ID="Owner" />
<owl:Class rdf:ID="ConnectionType" />
<owl:Class rdf:ID="Location" />
+ <owl:ObjectProperty rdf:ID="dependOn">
+ <owl:ObjectProperty rdf:ID="hasProperties">
+ <owl:ObjectProperty rdf:ID="hasConnectionType">
+ <owl:ObjectProperty rdf:ID="useEncryptAlgorithm">
<swrl:Imp rdf:ID="Rule-1">
<swrl:body>
<swrl:AtomList>
<swrl:IndividualPropertyAtom>
<swrl:argument2>
<ConnectionType rdf:ID="GPRS" />
</swrl:argument2>
<swrl:argument1 rdf:resource="#PDA" />
<swrl:propertyPredicate
rdf:resource="#hasConnectionType" />
</swrl:IndividualPropertyAtom>
</swrl:AtomList>
</swrl:body>
<swrl:head>
<swrl:AtomList>
<swrl:IndividualPropertyAtom>
<swrl:argument2>
<Encryption rdf:ID="DES" />
</swrl:argument2>
<swrl:propertyPredicate
rdf:resource="#useEncryptAlgorithm" />
<swrl:argument1>
<swrl:Variable rdf:ID="#PDA" />
</swrl:argument1>
</swrl:IndividualPropertyAtom>
</swrl:AtomList>
</swrl:head>
</swrl:Imp>
<PDA rdf:ID="FSC-PDA" />
<ConnectionType rdf:ID="WLAN" />
<Encryption rdf:ID="RSA" />
We have implemented a prototype system, which can generate XACML5 policy based on the corporate knowl- edge. By using a predefined access control policy tem- plate with essential information of subjects and actions to be taken, XACML policy can be generated from the results of reasoning over corporate knowledge.
5. Related Work
WS-Policy [1] defines a framework and a model for the expression of the capabilities, requirements, and general characteristics of entities in an XML Web Services-based system as policies. Policy expressions allow for both simple
5OASIS eXtensible Access Control Markup Language, http://www.oasis-open.org
declarative assertions as well as more sophisticated condi- tional assertions. But the WS-Policy framework lacks for- mal semantics, it prevents us to determine its expressivity and computational properties.
KAoS Policy and Domain Services [4] use ontology con- cepts encoded in OWL to build policies. These policies con- strain allowable actions performed by actors which might be clients or agents. The applicability of the policy is defined by a class of situations whose definition can enclose compo- nents specifying required history, state and currently under- taken actions. Mandatory action can be annotated with dif- ferent constraints restricting possibilities of its fulfillment.
All of these approaches did not address knowledge ex- change and sharing among all the stakeholders in the e- business scenario.
Rein6is a decentralized framework for representing and reasoning over distributed policies in the Semantic Web.
Rein (Rei and N3) uses high level Rei concepts for poli- cies and N3 rules to connect these policies to each other and the Web. Rein allows policies to be represented in different policy ontologies and uses N3 rules, a semantic web rule language, for defining the connections in these networks.
Reasoning over these networks to obtain policy decisions is done using cwm, an N3 reasoner.
In our approach, the framework is somehow centralized, which is designed to support the special service platform and act as a policy management component.
6. Conclusion and Future Works
We have investigated a distributed knowledge manage- ment approach to help modeling web services policies. This approach is based on the concept of corporate knowledge through the use of VKCs. By integrating the knowledge management service, the semantic policy framework is able to access external VKCs which can provide application- specific knowledge on transactions in e-business. After the knowledge integration, rich corporate knowledge can be used to fulfill the task of reasoning and enrich the pol- icy with former unavailable information like non-functional requirement of service and business rules affected by the transaction processes. In the future, we will develop the prototype of the semantic policy framework further, espe- cially the part of knowledge management service. Two as- pects will be considered: trust of the agent [16] and policy negotiation with the help of corporate knowledge.
The aim is to enable a semantic policy framework and knowledge management methodology, which enable secu- rity, trust and QoS in service-oriented computing environ- ments and provide a novel solution for fields like e-business, telecommunication and enterprise application intergration.
6http://groups.csail.mit.edu/dig/2005/05/rein/
Acknowledgments
One of the authors, Dong Huang, would like to thank the team at Siemens CT for their guidance and support whilst conducting this research. In particular, he thanks Jorge Cuellar for his valuable contributions towards this work.
References
[1] S. Bajaj, D. Box, and D. Chappell. Web services policy framework (ws-policy). Technical report, IBM and BEA and Microsoft.
[2] S. Battle, A. Bernstein, H. Boley, M. Gruninger, and R. Hull.
Semantic web services framework (swsf) overview version 1.0. Semantic Web Services Initiative (SWSI), 2005.
[3] M. Bonifacio, P. Bouquet, and R. Cuel. Knowledge nodes:
the building blocks of a distributed approach to knowl- edge management. Journal of Universal Computer Science, 8(6):652–661, 2002.
[4] J. Bradshaw and A. Uszok. Representation and reasoning for daml-based policy and domain services in kaos and no- mads. In AAMAS ’03: Proceedings of the second interna- tional joint conference on Autonomous agents and multia- gent systems, pages 835–842, New York, NY, USA, 2003.
ACM Press.
[5] J. Calmet, P. Maret, and R. Endsuleit. Agent-oriented ab- straction. Revista (Real Academia de Ciencias, Serie A de Matematicas), 98(1):77–84, 2004. Special Issue on Sym- bolic Computation and Artificial Intelligence.
[6] O. S. Coalition. Owl-s: Semantic markup for web services.
Whitepaper Version 1.0., 2004.
[7] N. Damianou, A. Bandara, M. Sloman, and E. Lupu. A sur- vey of policy specification approaches. Technical report, Im- perial College, 2002.
[8] K. Decker, M. Williamson, and K. Sycara. Matchmaking and brokering. In Proceedings of the Second International Conference in Multi-Agent Systems(ICMAS’96), 1996.
[9] D. Fensel and C. Bussler. The web service modeling frame- work wsmf. Electronic Commerce: Research and Applica- tions, 2(1):113–137, May 2002.
[10] G. Fischer and J. Ostwald. Knowledge management: Prob- lems, promises, realities, and challenges. IEEE Intelligent Systems, 16(1):60–72, 2001.
[11] M. Hammond. Virtual knowledge communities for dis- tributed knowledge management: A multi-agent-based ap- proach using jade. Master’s thesis, University of Karlsruhe and Imperial College London, 2004.
[12] J. Hendler. Agents and the semantic web. Intelligent Sys- tems, 21(2):30–37, 2001.
[13] I. Horrocks and P. F. Patel-Schneider. Swrl: A semantic web rule language combining owl and ruleml. Technical report, The Rule Markup Initiative, May 2004.
[14] D. Huang. Semantic description of web services security contraints. to appear in Proceedings of the Second IEEE International Conference on Service-Oriented System Engi- neering (SOSE2006), Shanghai, China, October 2006.
[15] D. Huang. Semantic policy-based security framework for business processes. In L. Kagal, T. Finin, and J. Hendler, editors, Proceedings of the Semantic Web and Policy Work- shop (ICSW05), Galway, Ireland, November 2005.
[16] D. Huang and S. Bracher. Towards evidence-based trust bro- kering. In Proceedings of the SECOVAL Workshop, held in conjunction with the 1st IEEE/CREATE-NET SecureComm, Athens, Greece, September 2005.
[17] M. N. Huhns, M. P. Singh, and M. Burstein. Research di- rections for service-oriented multiagent systems. Internet Computing, 9(6):65–70, November/December 2005.
[18] E. C. Lupu and M. Sloman. Conflicts in policy-based distrib- uted systems management. IEEE Transactions on Software Engineering, 25(6):852–869, November/December 1999.
[19] P. Maret and J. Calmet. Modeling corporate knowledge within the agent oriented abstraction. In Proc. International Conference on Cyberworlds (CW’04), pages 224–231. IEEE Computer Society, 2004.
[20] P. Maret, M. Hammond, and J. Calmet. Virtual knowledge communities for corporate knowledge issues. In Proceed- ings of 5th International Workshop on Engineering Societies in the Agents World (ESAW’04), volume 3451 of Lecture Notes in Computer Science, pages 33–44, Toulouse, France, October 22-24 2004. Springer.
[21] D. L. McGuinness and F. van Harmelen. Owl web ontology language overview. Technical report, W3C Recommenda- tion, February 2004.
[22] B. Motik, U. Sattler, and R. Studer. Query answering for owl-dl with rules. In Proceedings of International Semantic Web Conference 2004, pages 549–563, 2004.
[23] J. OSullivan, D. Edmond, and A. H. M. ter Hofstede. Formal description of non-functional service properties, queensland university of technology. 2005.
[24] P. F. Patel-Schneider. What is owl(and why should i care)?
Invited paper for the Ninth International Conference on the Principles of Knowledge Representation and Reason- ing,Whistler, Canada, 2004.
[25] M. Sloman. Policy driven management for distributed sys- tems. Journal of Network and Systems Management, 2:333–
360, 1994.
[26] I. Toma and D. Foxvog. Non-functional properties in web services. Technical report, DERI, 2006.
[27] C. D. Walton. Uniting agents and web services. AgentLink News, (18):26–28, August 2005.
[28] M. Weber. Economy and society. University of California Press, 1986.
[29] G. Weiss. Multiagent Systems: A Modern Approach to Dis- tributed Artificial Intelligence. MIT Press, 1999.
[30] M. Wooldridge. Intelligent agents. In W. Gerhard, editor, Multiagent Systems: A Modern Approach to Distributed Ar- tificial Intelligence, chapter 1, pages 27–78. The MIT Press, 1999.
