Service Discovery and Planning
written by gunther gerlach-2009
Current provisions for discovery are based on keyword searches through repositories. Keywords are nominated by service providers through publication and advertising features of software as a service (SaaS) functions. Details of message inputs, outputs, and methods are also captured from WSDL file scans and factored into searches.
Such discovery techniques are suitable in tightly coupled and well-scoped domains where service consumers can determine what services offer and how they can be independently utilized from search results. In other words, users are expected to know what they want before they search.
However, within the setting of more widespread web service ecosystems involving greater heterogeneity, this assumption breaks down. The wider the domain of an ecosystem – like an Amazon marketplace – the more general search schemas are. Therefore, the greater onus is on suppliers of services to enrich service descriptions that can be queried for the different variety of contexts in which services can be used. The current schemes of service classification on offer are noticeably simple by real-world standards. Consider for instance that only basic details of nonfunctional properties of services are available through current schemes.
Even if sophisticated schemes were available, as envisioned in ontology-based service semantics of Semantic Web technologies, it is doubtful whether overarching committees can reliably arbitrate service descriptions with sufficient foresight of the possibilities in which services would be utilized.
Committees have limited insights and their utility, paradoxically, lies in securing common-denominator consensus for ontological terms and references.
Left by the wayside are large sources of textual documentation about services, in regard to their business strategic, tactical, legal, or legislative, communal, jurisdictional, and demographic contexts – just to name a few possible types of service-related documentation. Sources of service knowledge dispersed through the environment in which services operate – jurisdictions, business missions, consumption points, and so on – can open up the variety of known and unknown contexts of services (a term known as latency semantics in the field of cognitive science).
In service ecosystems, this knowledge could be used outside the traditional bases of services, to determine how services could be procured through the service supply and distribution networks. Service providers could determine how to repurpose their applications in the variety of marketplaces.
For instance, more competitive packaging of application components through service endpoints might be determined. Similarly, service brokers could determine through which marketplace channels to target services, how demand could be driven up, what incentives to adopt, and so on.
To enable web service ecosystems, more suitable free-text search techniques would provide a strong competitive advantage. Whereas current ecosystems are expected to use UDDI-style repositories with keywords-based search, there is great potential in providing a suitable combination of free-text search techniques with ontology-based search techniques. This would furnish web service ecosystems with structured discovery as well as unstructured information retrieval style searches typical in service planning. Service planning – an earlier and more iterative phase of conventional service discovery – implies uncertainty of goal in conducting searches and a search agenda that becomes clarified with each search iteration.
Published by. BPTrends November 2005
