Catálogo de publicaciones - libros

We present four constructions for standard equipment which can be generated for every inductive datatype: case analysis, structural recursion, no confusion, acyclicity. Our constructions follow a two-level approach—they require less work than the standard techniques which inspired them [11,8]. Moreover, given a suitably heterogeneous notion of equality, they extend without difficulty to inductive families of datatypes. These constructions are vital components of the translation from dependently typed programs in pattern matching style [7] to the equivalent programs expressed in terms of induction principles [21] and as such play a crucial behind-the-scenes rôle in Epigram [25].

In this chapter we have analyzed whether existing ontologies are suitable for our purposes, thus answering the Question II.1: We have inspected one of the earliest and most prominent Web service ontologies, viz., OWL-S, as well as our own initial ontology of software components. We conclude that both are a big step forward with design principles suitable also for our purposes. Their reuse is possible in principle. However, both ontologies exhibit shortcomings that stand in conflict with our goals of having a high-quality, reference and heavyweight ontology. Their problems are very common also in more recent efforts (some of them are discussed in the related work chapter). We further conclude that most of the problems could have been avoided if a foundational ontology had been used as a modelling basis. Thus, the remainder of this part designs a new management ontology on the basis of a foundational ontology.

In this chapter we have answered the Question III.1: We have opted for an application server, but have come to the conclusion that other platforms would benefit from semantic technology as well. The next question we have answered in this chapter is III.2: We have seen that there are many potential sources that allow the (semi) automatic obtaining of semantic descriptions. Therefore, the number of manually provided descriptions can be kept small. We have then moved on to design an ontology-based application server that supports the semantic management of components and services. The resulting architecture is rather generic but provides a number of components to support application development in the Semantic Web (as introduced in our scenario in Chapter 4, Section 1.1). The following chapter presents a possible implementation of this design.

In this chapter we have proposed the semantic management of software components and Web services that trades off between modelling and management efforts (Main Question I: ). The trade-off point has been approached by identifying a set of use cases. Each of them responded to the Questions I.1 , I.2 , and I.3 The use cases propose the facilitation of some typical management tasks by a justifiable modelling efforts. The modelling requirements of the use cases also give us clear indications of what concepts a suitable management ontology must contain (Question I.4: ). The organization of these concepts in an appropriate management ontology is the subject of Part II.

In this chapter we have analyzed whether existing ontologies are suitable for our purposes, thus answering the Question II.1: We have inspected one of the earliest and most prominent Web service ontologies, viz., OWL-S, as well as our own initial ontology of software components. We conclude that both are a big step forward with design principles suitable also for our purposes. Their reuse is possible in principle. However, both ontologies exhibit shortcomings that stand in conflict with our goals of having a high-quality, reference and heavyweight ontology. Their problems are very common also in more recent efforts (some of them are discussed in the related work chapter). We further conclude that most of the problems could have been avoided if a foundational ontology had been used as a modelling basis. Thus, the remainder of this part designs a new management ontology on the basis of a foundational ontology.

We present four constructions for standard equipment which can be generated for every inductive datatype: case analysis, structural recursion, no confusion, acyclicity. Our constructions follow a two-level approach—they require less work than the standard techniques which inspired them [11,8]. Moreover, given a suitably heterogeneous notion of equality, they extend without difficulty to inductive families of datatypes. These constructions are vital components of the translation from dependently typed programs in pattern matching style [7] to the equivalent programs expressed in terms of induction principles [21] and as such play a crucial behind-the-scenes rôle in Epigram [25].

In this chapter we have analyzed whether existing ontologies are suitable for our purposes, thus answering the Question II.1: We have inspected one of the earliest and most prominent Web service ontologies, viz., OWL-S, as well as our own initial ontology of software components. We conclude that both are a big step forward with design principles suitable also for our purposes. Their reuse is possible in principle. However, both ontologies exhibit shortcomings that stand in conflict with our goals of having a high-quality, reference and heavyweight ontology. Their problems are very common also in more recent efforts (some of them are discussed in the related work chapter). We further conclude that most of the problems could have been avoided if a foundational ontology had been used as a modelling basis. Thus, the remainder of this part designs a new management ontology on the basis of a foundational ontology.

In this chapter we have answered the Question III.1: We have opted for an application server, but have come to the conclusion that other platforms would benefit from semantic technology as well. The next question we have answered in this chapter is III.2: We have seen that there are many potential sources that allow the (semi) automatic obtaining of semantic descriptions. Therefore, the number of manually provided descriptions can be kept small. We have then moved on to design an ontology-based application server that supports the semantic management of components and services. The resulting architecture is rather generic but provides a number of components to support application development in the Semantic Web (as introduced in our scenario in Chapter 4, Section 1.1). The following chapter presents a possible implementation of this design.

In this chapter we have responded to the Question III.3: We have reused the open source application server and have leveraged the wealth of tools provided by the Karlsruhe Ontology and Semantic Web tool suite, [Maedche et al., 2003]. KAON’s inference engine, ontology store and ontology editor have been applied to semantically enhance JBoss. The result of this fruitful combination is called , whose usefulness for building Semantic Web applications has been demonstrated by an example. The example has shown that without the KAON SERVER, application development for the Semantic Web leads to a one-off effort of combining software modules without the possibility for much reuse and extensibility. An assessment of the benefits of semantic management, as well as details regarding the application of the management ontology, follow in Chapter 10.

In this chapter we have answered the Question III.4: by taking the following steps: () we have specialized the core concepts and associations of the management ontology to reflect the idiosyncracies of the KAON SERVER. () We have removed concepts and associations that were introduced merely for reference purposes and () we have adapted the axiomatization to the KAON language. The resulting and version of the management ontology is actually applied in the KAON SERVER and can be obtained from http://cos.ontoware.org.

Finally, we have assessed the benefits of semantic management thus answering the Cardinal Question from the Introduction: We have taken a qualitative approach for assessment by revisiting the use cases introduced in Chapter 4, Section 2 and comparing management and modelling efforts with and without semantic management. The assessment demonstrated that the rather modest modelling efforts are clearly outplayed by the savings in management efforts.