Machine Learning Technology Artificial Intelligence Technology Natural Language Processing Technology Semantic Web Technology Ontology Technology Reasoning Technology Knowledge Information Technology Collecting AI Conference Papers Digital Transformation Technology
In the previous article, I described Reasoning Web 2009. This issue of Reasoning Web is a series of summer schools on theoretical foundations, modern approaches, and practical solutions for reasoning in the Web of Semantics. This book is the tutorial note of the 6th school held from August 30 to September 3, 2010.
of the sixth school in the series, held from August 30 to September 3, 2010.
This year, the school focused on applications of semantic technologies in software engineering and the reasoning technologies appropriate for such an endeavor. As it turns out, semantic technologies in software engineering are not so easily applied, and sev- eral issues must be resolved before software modeling can benefit from reasoning. First, reasoning has to be fast and scalable, since models and programs can be quite large and voluminous. Since many reasoning languages are exponential or NP-complete, approxi- mation, incrementalization, and other optimization techniques are extremely important. Second, software engineering needs to model software systems, in contrast to model- ing domains of the world. Thus, the modeling techniques are prescriptive rather than descriptive [1], which influences the way models are reasoned about. When a software system is modeled, its behavior is prescribed by the model, that is, “the truth is in the model” [2]; when a domain of the world is described, its behavior cannot be prescribed, only described by the model (“the truth is in the world”). Therefore, reasoning has to distinguish between prescriptiveness and descriptiveness, leading to different assump- tions about the closeness or openness of the world (closed-world assumption, CWA vs. open-world assumption, OWA). Third, while software modeling languages usually conform to a 4-level metalevel hierarchy (with objects on level M0, models on M1, metamodels on M2, metalanguages on M3), ontology languages usually only distin- guish TBox and ABox. Different metalanguages are used, different repositories, and different strategies for integration with application code. Basically, ontology and soft- ware modeling worlds are two technological spaces [3], and these spaces have to be bridged on each of the M0-M3 levels. Since bridging often requires transformations of models to ontologies and ontologies to models, flexible glue technologies are looked
for that hide the transformations from software developers.
These three requirements form only the top of the iceberg. There are many more
problems below the surface, and the lecturers of the school attempt to provide answers for at least the following questions:
– How can we limit the complexity of reasoning to polynomial time? For this purpose, expressive description logics have been developed and transfered to OWL- EL, a profile for OWL ensuring polynomial complexity. Many of these contributions were achieved here in Dresden, and Anni-Yasmin Turhan from the Description Logic group presented them in the tutorial “Reasoning and Explana- tion in EL and in Expressive Description Logics.”
– How can we reason with both rules and ontologies (hybrid reasoning)? Are we able to reason in an integrated fashion about domain models (which can easily be described by ontologies), requirements specifications (which talk about issues like business rules), and architecture specifications (which talk about rules for architec- tural styles)? What happens if a business rule from the requirement specification accesses the domain ontology? These fundamental questions of hybrid reasoning were taken up in the lecture on “Hybrid Reasoning with Non-Monotonic Rules” by Włodzimierz Drabent from the Polish Academy of Sciences and Linko ̈ping Uni- versity.
– Howcanontologiesbeembeddedintothemodel-drivensoftwaredevelopmentpro- cess (MDSD)? How can software models be checked on additional constraints? Can we model in an integrated fashion, that is, model in software languages while adding constraints from an ontology language? Which bridging technologies exist? These questions were discussed in the tutorial “Model-Driven Engineering with Ontology Technologies” by Steffen Staab, Tobias Walter, Gerd Gro ̈ner, and Fer- nando Silva Parreiras from the University of Koblenz-Landau.
– Can we use other techniques to speed up reasoning, such as approximation, in- crementalization, or database optimization technology? How do we integrate re- quirements ontologies (which talk about the domains of the world with OWA) with system architecture specifications (which talk about systems with CWA)? Jeff Z. Pan from the University of Aberdeen discussed these issues in his talk “Scalable DL Reasoning” [4].
– Not only data definitions and their languages, but also query languages have to be bridged. Problems like OWA/CWA need to be taken into account while querying syntactic (model) data and semantic data. An example for a bridging technology for two leading query languages from the ontology and modeling worlds (SPARQL and GReQL) was presented in the tutorial “Bridging Query Languages in Semantic and Graph Technologies” by Hannes Schwarz and Ju ̈rgen Ebert from the University of Koblenz-Landau.
Apart from these more conceptual tutorials, the school featured a strong discussion of application areas in software engineering.
– JensLemcke,TirdadRahmani,andAndreasFriesenfromSAPResearchKarlsruhe reported on an application of ontologies in “Semantic Business Process Engineer- ing.” SAP has developed a reasoning-based refinement method for business process specifications, in which a concrete, executable workflow can be shown to be a be- havioral refinement of an abstract business process. This method is very useful for enterprise process modeling, which is a focus area of SAP.
– Krzysztof Miksa, Pawel Sabina, and Marek Kasztelnik from Comarch (Krakow) showed how ontologies can be applied for the checking of constraints in domain- specific models in network device configuration (“Combining Ontologies with Do- main Specific Languages: A Case Study from Network Configuration Software”). Comarch is one of the leading providers for telecommunication network software and needs to control huge domain models.
– Michael Schroeder from Technische Universita ̈t Dresden presented “Semantic Search Engines.” Transinsight, his start-up company, has successfully built
several semantic search machines for different biomedical applications, for instance, GoWeb [5].
– There were two smaller tutorials on “Semantic Service Engineering in the TEXO
and Aletheia Projects” (SAP Dresden, Ralf Ackermann) and “Semantic Wikis” (So ̈ren Auer, Leipzig University).
We believe that the application of reasoning technologies in software engineering will be a fruitful field in the future. The summer school clearly showed that a number of challenges exist, but also that they can be overcome with an appropriate bridging technology.
We would like to thank all lecturers of the Reasoning Web Summer School 2010 for their interesting and inspiring tutorials. We also thank the Program Committee members for executing their reviewing duties, in particular under the given time constraints in spring 2010. Many thanks go to Christiane Wagner and Gundula Gelfert from CMD Dresden for the organization of the school. Thanks also to Steffen Ho ̈lldobler and the summer school “Computational Logic,” running in parallel with Reasoning Web this year [6], for their openness to share social activities. Finally, we also thank our funding agencies: this volume has been supported by the European Commission and by the Swiss Federal Office for Education and Science within the 7th Framework Programme project MOST number 216691.
Description Logics (DLs) are the formalism underlying the standard web ontology language OWL 2. DLs have formal semantics which are the basis for powerful reasoning services. In this paper, we introduce the basic notions of DLs and the techniques that realize subsumption—the fundamental reasoning service of DL systems. We discuss two reasoning methods for this service: the tableau method for expressive DLs such as ALC and the completion method for the light-weight DL EL. We also present methods for generating explanations for computed subsumption relationships in these two DLs.
This is an introduction to integrating logic programs with first order theories. The main motivation are the needs of Semantic Web to combine reasoning based on rule systems with that based on Description Logics (DL). We focus on approaches which are able to re-use existing reasoners (for DL and for rule systems). A central issue of this paper is non-monotonic reasoning, which is possibly the main feature of rule based reasoning absent in DL.
We discuss the main approaches to non-monotonic reasoning in logic programming. Then we show and classify various ways of integrating them with first order theories. We argue that for practical purposes none of the approaches seems sufficient, and an approach combining the fea- tures of so-called tight and loose coupling is needed.
Ontologies constitute formal models of some aspect of the world that may be used for drawing interesting logical conclusions even for large models. Software models capture relevant characteristics of a software artifact to be developed, yet, most often these software models have limited formal semantics, or the underlying (often graphical) soft- ware language varies from case to case in a way that makes it hard if not impossible to fix its semantics. In this contribution, we survey the use of ontology technologies for software modeling in order to carry over advantages from ontology technologies to the software modeling domain. It will turn out that ontology-based metamodels constitute a core means for exploiting expressive ontology reasoning in the software modeling do- main while remaining flexible enough to accommodate varying needs of software modelers.
One of the important aspects of Model-Driven Engineering (MDE) is to consider application-domain variability, which leads to cre- ation of Domain Specific Languages (DSL). As with DSLs models are concise, easy to understand and maintain, this approach greatly increases the productivity and software quality. Usually, the DSLs in MDE are de- scribed with a metamodel and a concrete syntax definition. The models expressed in the DSL are linguistic instantiations of the language concepts found in the metamodel.
However, for some of the application domains it’s not enough to con- sider the linguistic dimension of the instantiation. The problem arises when the domain itself contains the aspect of typing. This leads to an- other view on instantiation, called ontological instantiation . Since both aspects are present simultaneously, we refer to the combined approach with the term “two-dimensional metamodelling”.
In the following, we will exemplify the problem with a case study based on a real challenge found in the domain of network management. The solution we propose benefits from ontology technology which is ap- plied to enforce the semantics of ontological instantiation. Our approach presents significant differences comparing to the existing 2D metamodelling solution, although the motivations are similar. Thus, we consider our work as a case study of applying ontology enabled software engineer- ing in the area of DSL engineering, rather than a new metamodelling technology or an application of existing 2D metamodelling architecture.
The article is a result of joint work of the MOST project partners, applied within the case study provided by Comarch.
Software systems and software development itself make use of various technological spaces, e.g., relational databases, XML technology, ontologies, and programming languages. The usage of several technological spaces in a single system or development process requires a proper bridging between them. “Bridging” can be achieved by trans- forming concepts of one space into another, by using an adapter to make concepts of one space usable in another, or even by integrating both spaces into a single one, for instance.
This paper presents a transformation-based bridge between the query languages SPARQL and GReQL, with SPARQL originating from the semantic technological space and GReQL representing the model-based space. Since transformation of queries requires the prior mapping of the queried knowledge base, the approach also involves transforming the underlying data in the form of RDF documents and TGraphs, respectively. The benefits of the bridge are shown by applying it for traceability.
In this tutorial, we compare OWL-DL reasoning and Petrinet analysis for validating refinement and grounding of business processes.
(1) Process refinement: Like in software engineering, the implementation of a business process involves different interacting roles, such as business expert, analyst, process architect, and developer. Each role designs and refines different abstractions of the process until it is sufficiently refined. It is important to verify that the process models of the different abstractions are consistent.
(2) Process grounding: A sufficiently refined process has to be mapped on existing IT systems. Ideally, IT systems consist of components with a semantic an- notation of their behavior. The most specific process must respect all IT systems’ behaviors. Formally capturing process semantics enables to check automatically for consistent process refinement and grounding.
The classic application of semantic techniques in the area of static models is well understood. The analysis of business processes deals with dynamics. Model- ing dynamics is a challenge for current approaches of semantic Web services. We compare advantages and shortcomings of Petri net analysis and description logic (DL) reasoning for refinement and grounding validation.
In the next article, we will discuss Reasoning Web 2011.
コメント