Web Science is a relatively recent scientific field that entirely focuses on the study of the Web as a whole, and the macro effects it creates, for example with respect to social interactions, user participation and cooperation, societal impact, policy making, data and service re-use and integration, etc. Such effects are seldom engineered; instead, they emerge in the large-scale setting of the Web. Nevertheless, Web engineers should aim to design and develop Web applications that, through their intrinsic design features, form a solid foundation for the future evolution and development of the Web, and facilitate, as much as possible, desirable macro effects. One key ingredient for the future Web is Semantic Web technology. Using it not only provides the means of unambiguously and uniformly describing online and real world entities and the relations between them, it also allows to effectively link data from different sources, i.e. commonly called linked data.
The SEMARI project is set in the context of Web Science, and aims to develop methods and techniques for designing and building state-of-the-art Web applications in a sound and systematic way, that exhibit all feature commonly expected from modern Rich Internet Applications (i.e., better responsiveness, interaction capabilities and a rich user experience) on one hand, and make their data explicitly and semantically available for sharing and (re-)use on the other hand. The work is set within an existing Web engineering method, the object-oriented web solutions (OOWS) method, but strives to develop method-independent solutions where possible.
As such, the objectives of the research are threefold:
(i) to re-design and extend OOWS with the necessary models and conceptual primitives to design and generate modern Rich Internet Applications,
(ii) to extend OOWS to allow generating semantically annotated content on one hand, and allow (re-)using existing semantic descriptions on the other hand,
(iii) to demonstrate the usefulness of such semantic descriptions by supporting semantics-based context-awareness.
The work is demonstrated and validated using a case study of a RIA in the bio-informatics field.
To chart the research field, and identify under-investigated research topics, a systematic mapping study was performed on Rich Internet Application research done in the last ten years. Among others, the study showed that there is a relative lack of requirements engineering research, and topics as ‘mobility and device-dependency’ and ‘adaptation, personalization and contextualization’ are under-represented; more conclusions are available.
According to the lacks identified in the systematic mapping study, and in line with the objectives of this research project, a goal-oriented requirements engineering approach was extended with support for RIAs. A method for balancing non-functional requirements based on Pareto efficiency, also applicable for client-server configurations in the context of RIAs, was proposed and experimentally evaluated. This approach is compatible with OOWS, but also generally applicable.
Specifically to OOWS, the following extensions were made:
-the necessary models extensions to support Rich Internet Application features, such as event-handling, more advanced interaction scenarios and advanced presentation elements.
-the necessary models to represent the contents semantics based on existing ontologies, and representable in different formats (i.e., RDFa, microformats and microdata).
-a model to represent HTML5 rich presentation elements, and a configurable, default presentation mapping, which allows for rapid prototyping based on conceptual models.
-a prototype code generation engine for HTML5 and jQuery, to obtain state-of-the-art, semantically annotated Rich Internet Applications.
In the context of life sciences, more particularly a bio-informatics start-up, the developed models, methods and techniques were tested at hand of a case study. A study was undertaken to examine the applicability of OOWS for real-life, large-scale Web applications, which led iterative improvements in the proposed solutions, but also provides interesting conclusions and recommendations for the Web engineering community.
To demonstrate the usefulness of semantically annotated Web applications, and at the same time tackle an identified lack in RIA research, a method was developed to exploit RDFa annotations for context-awareness. This method is generally applicable, and combines a context-aware mobile framework, SCOUT, to gather and maintain context data, and client-side adaptation technologies to add context-aware features on the fly in existing web pages. In this line of research, important contributions were also made to gather, maintain and query small semantic sources (e.g., annotated websites) taking into account the limited capabilities of mobile devices.
All results were made available to the scientific and professional community through scientific publications in international journal and conferences. As such, this research helped to advance the state-of-the-art in Web engineering and Web science, and helps to bring more qualitative, multi-device Web applications to the general public. In summary, several methods and techniques to facilitate and ease development of complex, rich semantically annotated Web applications were contributed. The usefulness of semantically annotated Web pages was also shown, by deploying them to enable context-awareness. Finally, this research advanced technologies to maintain and handle semantic descriptions on a mobile device.
Óscar Pastor López
Period: 2010 -2012