This group develops concepts and techniques to cope with heterogeneity of service specifications for the success of worldwide service markets. Examples of existing service markets for software are such marketplaces as Eclipse marketplace or Google Play Store. In comparison to the state of the art, we work on the vision of a worldwide service market, in which service providers can offer their software services independent from a certain platform or service specification language. For that, we develop methods to perform such operations on services like matching or composition based on heterogeneous service specifications provided in a service market. This research is conducted in the scope of the subproject B1 of the Collaborative Reseach Centre 901 "On-The-Fly Computing" supported by the German Research Foundation (DFG).

Members: Svetlana Arifulina, Gregor Engels

Contact person: Svetlana Arifulina

Description

In the vision the On-The-Fly Computing, customer-specific software is developed on demand from distributed software, platform, and infrastructure services available on a worldwide market. Having requirements on a desirable service from a market customer, services have to be automatically discovered and composed in a service composition fulfilling these requirements. For that purpose, services have to be matched based on their specifications. For accurate matching, services have to be described comprehensively including their structural, behavioral and non-functional properties. One challenge of out scenario is the fact that market customers and service providers may use an arbitrary domain-specific languages (DSLs) to describe different properties of their services. Therefore, we investigate a parameterized comprehensive core language that covers different concepts needed to describe services for the accurate service matching. Concrete DSLs can be transformed into the core language, and specifications written in these DSLs can be translated into the core language for matching. For that purpose, we develop a user-friendly by-example model transformation approach. Another challenge in a worldwide service market results from the fact that accurate matching of comprehensive service specifications requires a lot of computation effort. That makes the accurate matching often infeasible in a service market because of its inefficiency. Therefore, we investigate how the comprehensive core language can be configured based on the cheracteristics of a certain service market, in order to perform as accurate matching as possible with a feasible efficiency. To sum up, parameterized service specifications serve as a basis to cope with heterogeneity of service specifications and to enable automatic service matching in a worldwide service market.

Research Challenges

In this area, we investigate the following research challeng

• Formal comprehensive service specifications using different domain-specific languages (DSL).
• Dealing with heterogeneous service specifications for service operations like service matching, service composition, and analysis of services.
• Language design for efficient and precise service matching.
• By-example derivation of semantic mappings between specification languages.
• Improvement of service specifications to improve their matching results.

We realize our concepts of parameterized service specifications in our Eclipse-based integrated tool suite called SeSAME (Service Specification, Analysis and Matching Environment).

SeSAME includes the following features:

• Creating comprehensive service specifications
  • Specification of different functional and non-functional service aspects based on a given language definition (Service Specification Language (SSL)) and its corresponding editors
• Matching of comprehensive service specifications
  • Service matching of functional and non-functional service aspects, which can be described using the SSL.
  • Single service specifications as well as test collections containing a set of service specifications can be matched.
  • An extensive view and reporting of matching results.
  • The approach of fuzzy matching of incomplete service specifications of different service aspects.
• Matching validation
  • Given a set of service specification pairs and the expected result of their matching, matching in SSE computes the so-called actual matching result for each pair of service specifications.
  • SSE can compute a matching statistics regarding such metrics as matching precision based on the comparison of the actual and the expected matching results.
• Transformation of service specifications written in other languages into SSL for matching
  • Hand-written transformations from SAWSDL, OWL-S, and UML into the SSL.
  • The approach of model transformations by-example to transform any specification language into the SSL (work in progress).
• Metamodel operations (applicable for every other metamodel-based language as well)
  • Calculating metamodel coverage of the SSL by certain service specifications
  • Creation of metamodel views on SSL by filtering out irrelevant language concepts in a view, in order to work only with the relevant ones
• Example service specifications and their expected matching results for different application domains
  • Touristic domain
  • University management domain
  • Water net optimization domain
  • Public sector domain

For more information, refer to the SeSAME website.

Project group AppSolut - "A Framework for Composed App Solutions" aimed at developing a framework for the composition of Android apps. The tasks include the development of a specification language to describe Android apps comprehensively. Furthermore, a matching concept has to be developed, in order to find suitable apps and to compose them correctly.

The students of the project group provided appropriate tool support for the framework - the App Composer Tool. Students presented the developed concepts and their realization at the "Software Engineering" conference 2014 in Kiel.

For more details on the results of the project group please refer to the AppSolut website.

Research topic in our reseach group relevant to this one are:

• Dynamic Meta Modeling
• Ontologies
• Service Oriented Architecture