Domain-specific Mashups: From All to All You Need

Stefano Soi Marcos Baez

Last years, aside the proliferation of Web 2.0, we assisted to the drastic growth of the mashup market. An increasing number of different mashup solutions and platforms emerged, some focusing on data integration (a la Yahoo! Pipes), others on user interface (UI) integration and some trying to integrate both UI and data. Most of proposed solutions have a common characteristic: they aim at providing non-programmers with a flexible and intuitive general-purpose development environment. While these generic environments could be useful for web users to develop simple applications, they are often too generic to address domain-specific needs and to allow users to develop real-life complex applications. In particular, proposed mashup mechanisms do not reflect those specific concepts that are proper of a given domain, which domain-experts are familiar with and could autonomously manage. We argue the need for domain-specific mashup architectures, also going beyond today's enterprise platforms, in which standard mashup mechanisms and components are driven by an underlying domain-specific layer. This layer will provide a service and component ecosystem built upon a shared and uniform conceptual model specific for the given domain. This way, domain experts will be provided with mashup components and mechanisms, following those well-known concepts and rules proper of the domain they belong to, that they are able to understand, use and, finally, profitably compose. In this paper, we will show the necessity of such an architecture through a reallife use case in the context of scientific publications and journals.