Citation Context

...ited number. Instead of trying to cope with these service styles through the application of semantic descriptions, we have, instead, used an extensible workflow enactment environment, called Freefluo =-=[11]-=-. The interaction with the web service is handled through a Java interface or processor. While this framework is not fully generic, it appears to answer our requirements; support for a new style of se...

Citation Context

...s at query time enables considerable architectural simplicity at this point. The Feta Engine is essentially a set of canned RDQL queries accessible via a web services interface. We currently use Jena =-=[5]-=- as our implementation backend as its query engine provides support for RDF(S) entailment. The canned queries that we currently support include: – An operation that accepts input of a given semantic t...

Citation Context

... selection from 1000 services to approximately 10 from which the user can then choose. Other authors have previously noted that this semi-automatic approach simplifies the task of service composition =-=[13]-=-, although they see this as a step toward further automation, rather than a strong user requirement. The opacity of the messaging structures means that descriptions do not have to relate to the intern...

Citation Context

...ions help agents (whether human or machine) interact with the service during its life cycle including discovery, composition, execution and monitoring. Considerable work on technologies such as OWL-S =-=[8]-=- and, more recently, WSMO [1], have focused largely on descriptions to enable automated composition of services. The requirement for full automation, and transparency of composition from the user pers...

Citation Context

...Grid project has built a domain ontology for describing bioinformatics services, described services semantically using that ontology, and used description logic reasoning to support service discovery =-=[20]-=-. However, we found this approach bought considerable complexity to the architecture, without addressing the requirements of our target users (biologists and bioinformaticians). This complexity arose ...

Citation Context

...rvices are hard to find; interoperability is poor. The primary mechanism for overcoming these problems has been the expert biologist; cutting and pasting between web delivered forms has been the norm =-=[15]-=-. Automated tools for the service composition have largely been built over the top of these web delivered services, often using Perl, and screen-scraping techniques. While this works, it tends to be f...

Citation Context

...erl, and screen-scraping techniques. While this works, it tends to be fragile to changes in the website. It is against this background that the my Grid project has operated. Along with other projects =-=[7]-=-, my Grid has developed more formal, programmatically accessible middleware, to enable transfer of information and composition of data and tool services into large workflows, addressing the real needs...

Citation Context

...ave levels of expressivity– as typified by the Semantic Web Layer Cake [4]. Much of the work on semantic web services has focused on the upper levels of the expressivity. In common with other authors =-=[9]-=-, we have found that “being light-weight and flexible trumps other features”. We believe that our XML and RDF(S) based architecture fulfils most of the requirements of the bioinformatics domain while ...

Citation Context

...from at least three areas: 1) Service Descriptions There are many aspects to a service which can be described at varying levels of detail, depending on the intended audience and use of the description=-=[19]-=-. The description needed to discover a service is different from that needed to configure a service, which may be different from that to invoke a service. Descriptions for unattended agents need to be...

Citation Context

...an query interface, many biologists are not. We would like to pursue further integration within Taverna to alleviate this need. In particular, we pursue the use of workflow context to filter services =-=[14]-=-. Hiding the explicit use of semantic service discovery should enable it to become a more natural part of the process of workflow building. 7 Discussion In this paper, we have described our applicatio...

Citation Context

...e to the internal structuring of the data; the best that we can do is describe the existence of a particular datatype 1 . However, the absence of formal structuring means that tools such as WSDL2OWLS =-=[12]-=- are of little use; there is little information in the WSDL file which can be mined from it. 4 Coping with Web Service Styles Of the 1000+ available external bioinformatics services available to my Gr...

Citation Context

...essaging format 2 for bioinformatics services [18]. This is not described further, but again, imposes additional semantics over normal web service invocation. 30% Web based REST services The Seqhound =-=[10]-=- sequence retrieval system delivers its services through a Representational State Transfer (REST) style interface, where all the information that is required for the service invocation is encoded in a...

Citation Context

...ervices, but exposes a stateful CORBA-like interface described later in this section. 30% Bio-Moby services The Bio-Moby project provides a registry and messaging format 2 for bioinformatics services =-=[18]-=-. This is not described further, but again, imposes additional semantics over normal web service invocation. 30% Web based REST services The Seqhound [10] sequence retrieval system delivers its servic...

Citation Context

...nformatics services. In a research domain where many workflows are developed in-house, experimented with, then either thrown away or extensively modified, speed of development is a rate limiting step =-=[17]-=-. There has been a large amount of interest in “Semantic Web Services”. In board outline, this augments standard Service Oriented Architecture with semantic descriptions of the services. These descri...

Citation Context

...g to apply these technologies to the bioinformatics domain. To this end, it has contributed to the generation of a rich service layer, has built a workflow engine and workflow development environment =-=[16]-=-. However, there is a difficulty. Even with a good environment, producing complex workflows is difficult, time-consuming and expensive. One of the reasons for this is the shear number of services whic...

Citation Context

...her than DL reasoning. Feta is meant as a light-weight semantic search engine rather than a full service registry, so this functionality is deferred to the standard web services registry, namely UDDI =-=[3]-=-. The core components can be grouped as semantic service publishing components (dark grey in Figure 2), service querying components (light grey) and the my Grid service ontology (unshaded) used by bot...

Citation Context

...upport is required for this process to encourage either external service providers, or service consumers to generate their own semantic service descriptions. To this end, we use the Pedro application =-=[6]-=-. This provides a GUI based interface which allows users to generate XML instance documents conformant to a given XML schema. The tool is also ontology aware and can provide easy access to the vocabul...

Citation Context

...enough utility for initial users of the system. In turn, this should encourage those in the domain to contribute new terms. This process has previously been used highly successfully in bioinformatics =-=[2]-=-. Due to the complexity of the domain, we choose to develop a complex property based ontology using OWL (initially DAML+OIL), which enabled us to take advantage of the reasoning at development time [2...