We are investigating the current Semantic Web Services approaches in order to select the most appropriate approach to build semantic internet TV services. Two major approaches are prominent which are top-down and bottom-up approaches, the former being heavyweight and supporting complex reasoning tasks and the latter being lightweight with limited reasoning capacity. We defend our assumption that semantic TV services require the strengths of both approaches and introduce a novel hybrid approach.
Nowadays, we share our knowledge via Wikipedia, our videos via Youtube, our pro les via Facebook, what we are doing via Twitter and our leisure time via Second Life. One of the last entertainment media, which endured resistant to cooperation and interactivity, was TV where the user was still a passive consumer. However the TV concept has evolved recently from the classic sofa TV to mobile and hybrid TV. This evolution has bene ted from the bandwith increase and smart phones wide spread use. Just like the Web content expansion phenomena, the evolution was coupled with the scattering of TV and multimedia content, where the user struggles to nd a relevant content, which perplexes the leisure time. Hence the need for new TV features : social, interactive, personalised and semantic. Such an ambitious idea requires many contributors to provide multimedia content, enrich the metadata with semantics, propose recommendation systems and build social networks around TV preferences. The NoTube1 project gathers many partners around the semantic TV idea. As a rst step towards collaboration, these partners expose their services via web services endpoints, that, in our vision, should be followed by a brokerage approach mainly aiming at the automation of Web service related tasks such as discovery, orchestration and mediation.
The actors in the semantic TV challenges, such as content producers, semantic enricher and recommender, have di erent backgrounds and concerns, which lead to a highly diversi ed state on choices of paradigms and technologies with three aspects of diversity: Di erent multimedia metadata schemas Content producers tend to maintain the format they are already using, as converting their whole multimedia base is expensive and need new expertise. Within NoTube, the main schemas being used are MPEG7, TV-Anytime2 and Electronic program guides (EPG) implying higher complexity on content processing.
Diverse tasks The exposed services achieve diversi ed tasks from content processing such as transcoding, scaling, cropping, to pro le analysis and metadata semantic enrichment.
Multiple services approaches Moreover, the partners expose their services
via multiple means such as Web API, REST [
Top-down approaches provide conceptual frameworks and languages, such as
OWL-S [
tools based on these formal SWS frameworks. For instance IRS-III is a
semantic execution environment that adopts the WSMO approach, videlicet that a
service description is expressed in terms of Goals, Mediators, Web Services and
Ontologies. IRS-III supports capability-based invocation: the request is a goal
to be achieved via the following intermediated operations [
Formalising these operations requires the use of rich knowledge representation languages and complex reasoners. However, the need to capture comprehensive and meaningful service semantics to allow reasoning-based automation of web services related tasks, contrast the requirement to lower the costs for providing services descriptions in order to simplify the modeling process for non SWS experts. 2.2
Based on the dictum that the top-down approach assumption, that the service
engineer describes semantics for the service before grounding these descriptions
to the services, is counter-intuitive, the bottom-up approach builds
incrementally upon existing Web services standards. Bottom-up or 'lightweight' SWS
approaches such as WSMO-Lite [
However, while those models are easier to produce [
But we enquire about : the real need of this automation, what are the most important tasks to be automated, their feasibility, the worth estimation compared 4
to the investment cost, the bene ts of an hybrid approach, could it maintain lower annotation costs while providing higher level of reasoning and how. The actors in the semantic TV challenges are rarely SWS experts, and therefore adopt the handy approach, namely Linked Services to annotate their services. However brokering such diversi ed and complex services as TV recommender and TV social services requires more exhaustive descriptions of the services and complex reasoning. Hence we aim to combine the strengths of both distinctive approaches into a coherent SWS framework. 5
As a rst step towards hybridization we investigated top-down and bottom-up approaches in order to predict the feasibility and cost of our approach. We sketch in Figure 1 a conceptual comparison between WSMO and MSM to evaluate the overlaps and di erences between both. 6
As we predicted there are a few overlaps, Web service and Ontology. WSMO high level concepts such as Goal and Mediator have no equivalent in MSM, likewise the Operation concept in MSM is omitted in WSMO as it uses more abstract concepts such as Choreography. However we dressed functional similarities in the properties that we drew close, such as has-condition and has-assumption, hase ect and has-postcondition. Besides many concepts model close notions such as Capability and Functional classi cation, Method and Choreography. Since the ambition for a possible interoperability between both approaches. 7 8
We will pursue our models investigation by a translation mechanism to allow IRS-III and iServe annotations interoperability.
In a highly complex and fragmented services pool such as the TV services, neither heavyweight nor lightweight semantic services approaches ful l service tasks automation, hence our motivation for a hybrid approach that provides handy services annotation and complex reasoning. Acknowledgments. We kindly thank the NoTube project for generously funding this research and being a validation framework for the discussed ideas.