Martin Junghans

Towards practical semantic web service discovery

Service orientation is a promising paradigm for offering and consuming functionalities within and across organizations. Ever increasing acceptance of service oriented architectures in combination with the acceptance of the Web as a platform for carrying out electronic business triggers a need for automated methods to find appropriate Web services. Various formalisms for discovery of semantically described services with varying expressivity and complexity have been proposed in the past. However, they are difficult to use since they apply the same formalisms to service descriptions and requests. Furthermore, an intersection-based matchmaking is insufficient to ensure applicability of Web services for a given request. In this paper we show that, although most of prior approaches provide a formal semantics, their pragmatics to describe requests is improper since it differs from the user intention. We introduce distinct formalisms to describe functionalities and service requests. We also provide the formal underpinning and implementation of a matching algorithm.

Integrating semantic Web services ranking mechanisms using a common preference model

Service ranking has been long-acknowledged to play a fundamental role in helping users to select the best offerings among services retrieved from a search request. There exist many ranking mechanisms, each one providing ad hoc preference models that offer different levels of expressiveness. Consequently, applying a single mechanism to a particular scenario constrains the user to define preferences based on that mechanisms facilities. Furthermore, a more flexible solution that uses several independent mechanisms will face interoperability issues because of the differences between preference models provided by each ranking mechanism. In order to overcome these issues, we propose a Preference-based Universal Ranking Integration (PURI) framework that enables the combination of several ranking mechanisms using a common, holistic preference model. Using PURI, different ranking mechanisms are seamlessly and transparently integrated, offering a single facade to define preferences using highly expressive facilities that are not only decoupled from the concrete mechanisms that perform the ranking process, but also allow to exploit synergies from the combination of integrated mechanisms. We also thoroughly present a particular application scenario in the SOA4All EU project and evaluate the benefits and applicability of PURI in further domains.

Behavior Classes for Specification and Search of Complex Services and Processes

In the Web there are a large number of (business) services with complex behavior, such as e-commerce Web sites that require multiple interactions with the user, as well as an increasing number of Web automation scripts to coordinate the execution of multiple complex services. However, while there are a quite a few search techniques for atomic Web services, search techniques for complex services are still rare and only foundational. In this paper, we present \textit{behavior classes} that have formal semantics as well as human comprehensible names in order to foster usability of specification of constraints, and efficiency of search for complex services and processes. Our approach enables automatic methods for (i) assigning behavior classes to complex behavior descriptions, (ii) checking consistency of such a classification, and (iii) computing behavior class hierarchies. Furthermore, human comprehensible names for the behavior classes increase usability by allowing for shorter service descriptions and requests. Our evaluation results show that a behavior class hierarchy can be exploited as an indexing structure to gain performance of search.

Empowering Business Users to Model and Execute Business Processes

Existing process modelling languages and especially executable process modelling languages are not designed for business users without programming knowledge. We therefore propose a novel Lightweight Process Modelling seeking to lower the entrance barrier for modelling executable processes. In this sense lightweight applies to the user interaction and means easy to understand in the context of the modelling language and easy to deploy, implement, and execute processes in a tooling context. Hence business users get advanced guidance during their modelling activities. This paper will provide a specification of a Lightweight Process Modelling process and the Language for Lightweight Process Modelling (LLPM). The LLPM formal semantic core is fairly rich, but it is designed to be rendered in a simple graphical form without undue loss of semantics. To achieve this we followed three design principles of lightweight modelling when supporting a business user: abstracting from executable process details, using semantic annotations, and reusing process parts through patterns and templates. In order to realize these design principles we have created new elements for the LLPM that are not yet implemented in existing process modelling languages. Selected concepts of existing process modelling languages like BPMN and BPEL complement the LLPM. In this paper we present a coherent specification of the elements, properties, and relationships. Further a design process is defined revealing the steps of enhancing the abstract graphical process models with execution details.

Efficient Search for Web Browsing Recipes

Information search in the Web can become cumbersome if the desired information is scattered across multiple websites. For instance, even though there exist pages listing track chairs of the past ICWS conferences and web accessible bibliography databases, compiling the list of recent journal publications of the ICWS track chairs with the help of existing search engines is still a time consuming task. It is even harder to find information from the Deep Web as it requires user interactions that are hard to simulate by automatic crawlers. Semantic search based on structured data aims at efficiently answering information needs but relies on the cooperation of providers to be able to access their data. We aim at providing an alternative solution by introducing web browsing recipes that are goal-oriented end user browsing processes containing instructions for accessing, extracting, and merging (dynamic) information from various websites. Browsing recipes can be shared and reused to allow users to benefit from the browsing efforts of others. In order to achieve this goal, the development of efficient search techniques is the main prerequisite for effectively sharing and reusing recipes. In this paper we propose an efficient search technique for finding browsing recipes from large recipe repositories. Our search technique necessitates a structured query specifying the required information along with constraints on the structure of the browsing processes. We augment explicit state representation based model checking technique by indexing structures tailored to the requirements of information search based on the recipes. The performance evaluation of our approach reveals the impact of the indexing structures on the overall recipe search efficiency.

Meaningful Service Classifications for Flexible Service Descriptions

We present a formal underpinning for \textit{Web service classes} by viewing them as a set of services that fulfill a logical combination of constraints on functional and non-functional properties. A hierarchy of service classes is automatically derived by their formal definition and can be exploited for an efficient service retrieval. In addition, we show in this paper how service classes can be used (i) to create service descriptions without specifying precise property values and (ii) to create service requests that can use service classes to express ranges of desired property values.

Swapping Out Coordination of Web Processes to the Web Browser

Consumption of (business) processes provided in form of Web sites have become a part of our daily life for attending our personal and business needs. In order to obtain the best solution for a particular task, users often combine several Web processes. However, the coordination of the execution of such Web process compositions is completely manual demanding the user to enter same or logical dependent data multiple times. We argue that a part of such coordination effort could be automatized by swapping out the uncreative coordination tasks to the Web browser. Our solution allows users to compose Web processes as generic solutions and execute the compositions with appropriate parameters every time they need to perform a concrete task, thus relieving them from a lot of manual coordination effort. We show how such Web process compositions can be formalized, obtained and executed inside a common Web browser with automatic flow of data among different parties despite heterogeneous data.

Towards simulation-based similarity of end user browsing processes

For increasingly sophisticated use cases an end user needs to extract, combine, and aggregate information from various (often dynamic) web pages from different websites. Current search engines do not focus on combining information from various web pages in order to answer the overall information need of the user. Semantic Web and Linked Data usually take a static view on the data and rely on providers cooperation. Web automation scripts, initially developed for testing websites, allow end users to capture their browsing activities as executable processes and share them with other end users. A script can contain instructions for accessing, extracting and merging (dynamic) information from various websites for a particular purpose. Techniques for allowing users to search for scripts that satisfy complex constraints restrict to existing scripts in the repository, i.e. they do not deduce scripts that may satisfy the request as well. In this paper, we show how semantic descriptions of web sites can be derived from such scripts, and how such semantic descriptions of web sites along with usage information present in the scripts can be used to obtain new scripts with similar functionality.