Confidential Draft for Distribution to Nessi Members Only

ed Infrastructure Service Modelling SVN Modeling Infrastructure Modelling SBS/SBA Modelling Formal Languages Context Modelling “ S y s te m ” P la tf o rm offered as services Business Process Execution Information Services Service Communication Service Discovery Mediation SLA Negotiation Service Composition BPM view SVN Lifecycle Management S e rv ic e P la tf o rm Service Coordination Integration Services Reasoning Service Execution Lifecycle Management Service Registration Business Process Modeling M o n it o ri n g “ S y s te m ” P la tf o rm S e rv ic e P la tf o rm Figure 2. Draft Skeleton Layered Functional Architecture. “Engineering tools” concerns the research area of needed in order to build (including modelling, development, simulation, etc.) services, service-based systems, etc. CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p47/99 10-Dec-07 Service Security Model Business Model (e.g. terms and conditions) Deployment Model Behavioural Model Performance Model Figure 3. Service non-functional aspects. Issue: • Who is responsible for monitoring management? If this is not coupled with the framework (as a single or central engine) this could be “simply” (only) coupled with the service provider, so the question is how to exchange monitoring information and to which extent they exist monitoring standards. “How to read this section”: the intent of this document is to be as much as possible agnostic with respect to specific approaches. Nevertheless, in the following sections, for some functionalities approach and abstract description are so much intertwined that it is difficult to separate them. 3.2 Description of Functionalities [Description of each envisioned functionality (each yellow block) at coarse-grained according to some predefined template covering the following aspects: • overall description, • intended users, • technological challenge or problem addressed, • positioning with respect to SofA (State of the Art), Gap to fill in, • Scientific and Technological approach proposed (different approaches can be provided, e.g. semantics, security, soa), • expected results (it includes benefits) • ...] CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p48/99 10-Dec-07 Hereafter we describe the functionalities that we do envision at the level of each coarse grained functional block of the layered view of the Open Service Platform as depicted in and resulting from the work achieved in section 4.1 (clustering of functionalities according to the Conceptual view of the Skeleton Architecture Sketch as promoted/pushed so far and which has been revised in the context of SRA Vol 2. production – see Section 6 for complete update also process attached to this update). 3.2.1 Service consumer layer Functional blocks attached to this layer are there to answer demands coming from service consumers. Thus they result from an in depth analysis of their (service consumers) specific requirements. They not only cope with existing and/or well known demands but also and probably more importantly to first acquire and second accommodate new service demands which may partly require to evolve existing services and partly to even come up with completely new breed of services. Last but not least each of those functionalities will be assessed against a number of key criteria such as usability, versatility, openness and user acceptance. Two major functional blocks have been identified here: • User-service Interaction • End-user Interface 3.2.1.1 User-Service Interaction The growing relevance of services in today’s economy increases the importance of highly effective and efficient user/service interaction leading to more productivity. Productivity can be increased particularly along with the following three phases of interaction. Retrieval (moving from data retrieval to service retrieval, Service Access, accessibility): In the Internet of today, seeking specific services can be a time-consuming task. With a significantly growing number of services in the Web, this complexity as well as the uncertainty with respect to exact service features and quality will even deteriorate the productivity of finding adequate services which exactly meet individual preferences and needs. ServiceUsage and Intelligibility: In the Internet of today, the utilization of electronic services involves difficulties when to cross-service interoperability and cross-device usability are required: Mostly, users intend to utilize a set of services and desire to arrange and compose them in a way that solves their unique problems. However, current technologies do not support seamless interaction and message exchange between different services. Also, services are hardly usable across different (mobile) devices without interruptions, again leading to decreased productivity of user/services interaction. (Implicit or Explicit) Intelligent Management: Managing services represents a third hurdle to efficient and effective user/service interaction. Services are rarely subject to formalized and standardized performance parameters which allow for manageability and thus predictability and dependability of the interaction. User will less and less in the near future directly deal with the technical implementation part of the services needed to answer their demand or request. Therefore User Service interaction would have to be mediated. With the help of the following means, the above discussed hurdles can be tackled adequately: First, intermediaries are required which take over institutional functionality. In a Web of countless different services, such entities are needed to ensure market transparency, navigation capabilities and a certain degree of governance. By gathering feedback (e.g., with respect to performance) about services, by building up a comprehensive services registry which simplifies service retrieval and by monitoring and enforcing previously agreed conditions of services utilization and bindings, uncertainty inherent to users/service interaction can be reduced significantly. The utilization of context is crucial as well. Both user context (knowledge, profile, preferences, idiom, association to social networks) and usage context (static and dynamic characteristics of devices used for accessing services, geographical location and time, connection bandwidth) shall be leveraged to improve service retrieval and CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p49/99 10-Dec-07 utilization. By considering all major context parameters, an appropriate service can be found faster than without contextual search. Also, depending on user and usage context, the utilization phase of user/service interaction can be enhanced: Depending on bandwidth and used device, for example, the visualization should be adapted. Usage of context should promote goal-oriented approaches. Referring to contextual knowledge about the user, environment should also empower end-users to clearly perceive how knowledge is put at their service at any time (captured from or provided by him, captured from context, shared with other users, during content and service discovery, service access, while tagging, etc.) enabling them to reach the highest degree of productivity during interaction. Third, besides of pulling information from the users to services (e.g., for contextual search), services shall also be pushed towards users proactively. Indeed, unsolicited pushing of services according to automated inference of current user needs may represent a paradigm shift and a major advancement in terms of user/service interaction. To achieve a more productive level of user/service interaction, Social Software is called to play a key role here in since it enables people to rendezvous, connect or collaborate and as such empowers them. One of the key concepts to consider here in Social Software is that of folksonomies 47 which have shown to be a powerful tool for associating metadata to resources and organizing knowledge in a distributed way. Folksonomies have their strength in their ease of use, which allows the generation of an increasing amount of valuable metadata by non-trained users. The power is inherent to users that massively feed the system with tags, this way avoiding the cost-intensive task of manual annotation by an expert and the inherent difficulty in developing technologies that perform automated annotation of audiovisual content. This way, the low-cost metadata provided through folksonomies can be exploited to enhance search engines and also to improve the navigation experience. Nevertheless relevant new areas of research have to be addressed in order to find out how folksonomies can be structured, linked to services (they are today mostly linked to contents) and combined with more formal technologies like ontologies in order to get the best of the two and better serve service access and management from a user-centric (goal-directed) perspective. To improve the problem of composing different services to create one user interface, mash-up technologies are required. The BusinessWeek 48 explains this novel form of using Web-based resources: “Mash-ups portend big changes for software companies, Web sites, and everyone online. No longer just a collection of pages, the Web is morphing into a sort of global operating system, à la Microsoft [...] Windows. And now, people are learning to program Web 2.0 with much of the same innovative energy of the personal computer's early days.“ With the help of such technologies, users are empowered to create their own interfaces, arrange services according to their unique preferences and also define message exchange patterns between different services (dependencies between services). 3.2.1.2 End-user interfaces In order to allow for realizing Mash-ups and leveraging the knowledge of the “long tail” through folksonomies, adequate end-user service interfaces are required. Service-oriented Architectures are particular tailored and used for business applications within one or across several enterprises. Services provide access to business data (e.g. data of a customer) and allow for managing business workflows. However, services are rarely directly used by end users. User interfaces together with complex control logic must be developed as an additional layer on top of services. This development step is currently only insufficiently supported by methodologies and tools. Currently, user interfaces are created independent from service interfaces and business processes and thus have to be manually designed. The separation of service design/ implementation and the actual user interface design leads to a gap between the developers’ and the actual users’ understanding of a service. An integrated approach for creating user interfaces for several services is not considered in research and Web Service standards. Even with proper user 47 Folksonomies are a collaborative, open-ended labelling system that enables users to categorize different resources such as web pages, on-line photographs, audio or video. 48 Mashing the Web, The Economist 376 (8444), Special section, p. 4, 2005 CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p50/99 10-Dec-07 interfaces for each individual service of a complex scenario, an integrated view on the whole functionality would not be possible. In consequence, each service would only be usable via its own user interface, resulting in user interface cuts between neighboured services. The integration of several services to a common user interface is missing. As an essential part of the NESSI initiative, a “face has to be put on services” so that these can easily be utilized by human users with only little IT literacy. Therefore, novel concepts are in the focus of the work, particularly with respect to service engineering methods, novel techniques for visual, model-driven service and service interface development, methods and solutions for adaptive user interfaces as well as runtime environments which support a close interrelationship between user interfaces and underlying services. From a consumer adaptation perspective the next generation SOA front-end envisioned here in the context of NESSI Open Service Platform will also offer to end-users the capability of context-aware discovering, selection, tagging, composing (including visual ones) and rendering of resources, following a self-service philosophy. This being mainly due to the fact that end-users can only deal with things they “can see” and “understand” (i.e. speak/has sense to them) at the user interface level (i.e. things that can be visually represented and made “intelligible”). Deeply related to service usage the dynamic (on-the-fly at runtime) mashing of Web services up to the mashing of Web sites/applications can be addressed using, extending and coupling Web 2.0 and semantic techniques (Mash ups and aggregators 49 are among the research areas for service usage). Such a front-end platform would operate as an open market-place where content/service providers and end-users get connected. An open market-place where SMEs and large companies may publish their offer, making it visible to endusers. A place where end-users can search/pay for and consume content and applications delivered as services, also a place where they could place specific demands using their (almost) own wording while being assisted in doing so. This is particularly true in the context of on-demand provisioning and delivery of services according to well defined service level agreements where a major challenge at Service customers level remains to first describe and second negotiate required service characteristics in a way that is meaningful to their level of expertise but also sufficiently accurate to be dealt with. At the end, users will get fully empowered, being able to select resources of their interest, compose and configure using piping/mash-up techniques their own services/business processes and application suites, while validating and assessing their performance beforehand (predictability based on behavioural serviced model encompassing both functional and non functional requirements as expressed) through test suites and visualization tools and at the end sharing these results as well as knowledge with other end-users in order to fine tune and capitalize on it. This personalization process being carried out in a visual manner so as prior IT expertise is no more required. 3.2.1.2.1 Personal Interface for the Information Society Present interface and desktop implementations not only not implement fully neither the desktop metaphor nor the document orientation: generally they avoid the easy integration of third party components. Present desktop software is not architected to direct Internet/Web interoperation being still needed the intermediation of a web browser which is getting itself into an own desktop software platform. SMEs need an interface to extract full power to the personal memory, to group work, group discussing and consensus reaching, to the creativity and an easy access to enterprise resources and applications. On the other hand SMEs need to reduce the complexity of creation and deployment of whole new types of applications based on extensive use of multimedia, remote services, knowledge management, information inference and mobility. 49 These are similar to mash-ups superficially, but are driven by a different goal. The primary goal of aggregating sites is to collect data from heterogeneous and multiple sources and republish the cleansed, integrated and aggregated data at a single point-of-access. These suffer from the well-known information integration problem which may be eased if semantic technologies are used to find content relating to the same instances, enabling mappings between different ontologies created by experts to be re-used by all aggregators. CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p51/99 10-Dec-07 The goal is an object oriented / service based / semantic powered graphical desktop with full support of identity, personal memory/knowledge, coordination between mobile devices and fixed and remotes ones, webarch based, refactoring of FLOSS desktop software, quality and maturity criteria for desktop applications and support for model MDA languages in order to reduce the risks and costs of development and deploying software and able to work as interface of the «executable enterprise» allowing the creation of powerful interfaces with similar or least costs than the present Rich Internet Applications alternatives. 3.2.1.3 Context Modelling [Hu06] defines context as follows: “The context of an idea or event is the general situation that relates to it and which helps it to be understood”. Systems theory [Bertalanffy68] deals with context as it aims to identify the boundary between systems and their respective environmental context. A widely accepted approach is to assume the boundary delimiting that part for which we seek a detailed understanding in terms of the constituent components, their behaviors and relationships. What is inside the boundary is "the system". The "context" or "environment" is everything that is outside the boundary, but may interact with the system and influence its behavior. Contextual computing aims at developing systems incorporating awareness of and the ability to adapt to contextual information about users and their respective environments with the goal to enhance application services. Currently, technical applications frequently behave the same way regardless of individual user preferences, situational parameters, applied technologies and others “contextual parameters”. In a nutshell: Contextual computing is about extending systems with the ability to detect contextual information, to represent it and display it to the user, to manipulate it, and to influence the behavior of applications with the goal to support human users in a proper and personalized fashion. While contextual computing research aims to develop systems incorporating awareness of and the ability to adapt to user context [Hu06], context modeling comprises the formal representation of contextual information (which can be considered as a basis for contextual computing). Early models mainly addressed the formalization of context with respect to one single application/ application class. Today, uniform context models, representation and query languages as well as reasoning algorithms that facilitate interoperability of applications and context sharing are in the focus of research [Strang04]. 3.2.1.3.1 Exemplary Use Cases of Contextual Computing: Numerous different promising use-cases of contextual computing have emerged and continue to emerge in various fields such as e-Business, e-Learning, Healthcare, Mobile Communication Networks and many more [SHC00, Crowley02, CCRP02, Crowley03]. AmbieSense 50 represents one example for the extensive utilization of contextual information for the adaptation of technical systems. The goal of AmbieSense is to provide personalized, contextsensitive information to mobile, “nomadic” user. An ambient information environment will be provided through a combination of context tag technology, a novel software platform to manage and deliver the relevant information and personal computing devices to which the information is served. The “MILK 51 ” project has been funded by the EU and developed context-aware applications for information sharing in distributed organizations. Another project dealing with related issues has been conducted by XEROX and is called WID 52 (Web Information Discovery). It aims at automating the “Hidden Web” discovery process and is divided into three subprojects: the discovery, the analysis, and the classification of resources. Lieberman at the MIT Media Lab [Liebermann00] analyzed the trade-off between the need for abstraction (to allow for modular and tractable systems) and the need for context-sensitivity. Several applications that demonstrate the power of context sensitivity have been presented at the MIT Media Labs. The use of contextual information has recently also been investigated in the field of communication networks 50 http://www.ambiesense.com/ 51 http://klee.cootech.disco.unimib.it/milk/ 52 http://www.xrce.xerox.com/competencies/contextual-computing/projects/wid/home.html CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p52/99 10-Dec-07 [ESKS06, KAESS06]: To improve information routing and dissemination, for example in the case of emerging vehicular ad-hoc networks, the use of manifold contextual information is currently being analyzed. 3.2.2 Consumer Adaptation (functional blocks) 3.2.2.1 Mapping users perspectives to business/Integration Open question: what do we plan to put here ? no slide available yet ? As a first attempt ... We could start here to mention here the systematic grounding of functional requirements vs. non functional requirements from the business level down to the physical infrastructure. A good example could be here the systematic grounding of SLA ... Service providers require the management of service level agreement (SLA) contracts in business terms, e.g. for minimizing financial penalties for service-level violations or maximizing service-level measurement based customer satisfaction metrics. A considerable numbers of languages (i.e. WSLA, WS-Agreement, WS-Policy) have been developed that are used for specifying SLAs between service providers and consumers. These languages are focused on the specification of functional and quality properties of web services. However, they do not provide support for specifying information that would be necessary in a business context. Early initiatives have been launched for developing an information model to represent business concepts such as SID (Shared Information/Data); see the TeleManagement Forum (TM Forum). To overcome this limitation one approach could consist in significantly enhancing the information model defined by the TMF in SID by including business SLA definition. Furthermore, the project will enable the service providers to apply to the whole business lifecycle in a consistent way. 3.2.2.2 Context handling Functionalities introduced hereafter are all related to context management at runtime (context modelling being addressed elsewhere in the document see section 3.2.1.3). Context-aware computing is traditionally concerned with the impact that contextual information has on the behaviour of mobile devices and software entities. This body of work has taken a somewhat narrow view of context, focusing on simple context parameters such as location, light and noise levels and deployment platform, which can be easily obtained in an automated fashion. A more recent development of this idea is being developed under the Wireless World Initiative, 53 (FP6). Context in all these projects is focused on the profiles of users, devices and communication channels that help to adapt applications, user interfaces and to a limited extent the delivery methods (e.g. bandwidth determines resolution). To facilitate the deployment of semantic services on a global scale, “context” will cover a range of aspects ranging from system and location information to social settings and legal regulations. In the context of NESSI and this section, we give to the term “context” the widest meaning that is possible, comprising both user context (knowledge, profile, preferences, idiom, info about social networks the user belongs to, etc.) and usage context (static and dynamic characteristics of the device used for access, geographical and time location, connection bandwidth, etc.) Context modelling will serve several purpose when managed at runtime. Among others we envision the following functionalities to be offered: • Context-based user service interaction: personalization of user-service interaction based on information/knowledge we have about the user (his/her profile, preferences, ...). Dynamic management of user context either to personalize/adapt the service interaction and/or learn from it to capture new knowledge and thus enrich/evolve the underlying model(s) (e.g. user model). With regard to the interaction 53 http://www.wireless-world-initiative.org/ CONFIDENTIAL DRAFT – FOR DISTRIBUTION TO NESSI MEMBERS ONLY SRA_Vol2_20071209_GA_2.doc p53/99 10-Dec-07 of users, one important question is to provide secure authentication between users and the front-end platform and also between users and service providers. One additional requirement is to deal with multiple identities of users (e.g. work and private roles). Besides, users tend to be nomadic, i.e. access services in very different environments using different devices. It has to be investigated where and how the context and profile information can be stored, retrieved and managed. Finally, another important issue to deal with, is privacy. • Context-aware services and processes: use of contextual information (semantically based) to enhance/advance services such as search and discovery services (e.g. context-aware discovery of services to find the “right” service among the billions of services offered) and/or service composition. ContextAwareness to support collaboration across business networks in context of business processes. Methodologies and tools to acquire service context from social networks. SLA-aware service management to include SLA awareness in dynamic service discovery & composition scenarios, both within and across domains which is to date clearly missing. Service Value Network processes aware of SVN policies, contracts and SLAs. Context management and adaptation will be mainly achieved by including context-dependent factors ranging from immediate concerns of location and language to legal issues and how financial regulations can be managed. Among others, usage of the following techniques and/or mechanisms, as well as of others which will emerge in the future, will be explored to handle contextual information and properly adapt either service and/or user interaction: • Context parameters (e.g. the Aspect-Scale-Context model behind CoOL) provided through the means of WSMO goals, • Context-specific rules and constraints (as enabled by WSML-Rule and RIF) to express policies and business rules, • Context-specific functionality, such as context adaptation involving service composition, where a core service is composed with a set of services providing context-specific information, • Methods and tools to extract context from tag clouds and informal folksonomies through means such as rich tags, OpenID, and FOAF. 3.2.2.2.1 Teaching Applications The adoption of a new application has not trivial costs as part of the ROI formulas. Similarly, the adoption of new practices of enhanced processes implies such costs. When combined, their impact could be, in the worst case, greater. This implies strong barriers to the productivity enhancement specially in knowledge based companies. We identify the need of a new way to conceive applications/interfacing/knowledge development and deployment simplifying it for converting the book into an application and the application into the reference book of the managed knowledge. Look for technologies for formal structuring and annotation of knowledge, processes and procedures allowing their execution and enough enhancements on interfacing and pedagogical active methodologies for making possible applications which teach themselves or book that executes their teachings creating new levels of auto-study. The goal is a set of new methodologies and tools to support the learning and acquisition of solutions built on business process modeled knowledge. This knowledge should be used as «a software solution» and as «digital enciclopædia», and managed in the user side with, preferably, native interface integration.