1. Hera Presentation Generator TU e Information Systems Group Department of Computer Science Eindhoven University of Technology Flavius Frasincar flaviusf@win.tue.nl flaviusf@win.tue.nl Geert-Jan Houben g.j.houben@tue.nl g.j.houben@tue.nl Peter Barna pbarna@win.tue.nl pbarna@win.tue.nl Information Systems Group Department of Computer Science Eindhoven University of Technology PO Box 513, NL-5600 MB Eindhoven, The Netherlands Hera project: http://wwwis.win.tue.nl/~herahttp://wwwis.win.tue.nl/~hera SEMANTIC WEB INFORMATION SYSTEMS (SWIS) Data-intensive systems Data integration from heterogeneous sources Presentation personalization based on user preferences and browsing platform Use of Semantic Web technologies HERA METHODOLOGY Hera is a model-driven design methodology for SWIS. It distinguishes the following phases: Data collection: makes available data coming from different possibly heterogeneous sources. Presentation generation: produces hypermedia presentations tailored to the user and the user’s browsing platform. Hera uses RDF(S), the foundation language of the Semantic Web, for model representations. PRESENTATION GENERATION PHASE The presentation generation phase of the Hera methodology is shown in Figure 1. It is based on the the following steps: Conceptual design: specifies the application data in a uniform manner independent from the input sources. The result of this activity is the conceptual model (CM). Application Design: defines the navigational structure used for data delivery. The result of this activity is the application model (AM). Presentation Design: specifies the look-and-feel aspects (e.g., layout, fonts) of the data presentation. The result of this activity is the presentation model (PM). Implementation: based on the previous design specifications a pipeline of data transformations produces a Web presentation. The implementation comes in two flavors: XSLT-based and Java-based. For the XSLT-based implementation, the XSLT stylesheets describe transformations between the RDF/XML serialization of the Hera RDF models. The design of the personalization of the presentation is distributed over all the previously identified design steps. The user/platform profile (UP) stores the user preferences and platform characteristics that influence the presentation before the user starts browsing. The UP is based on the CC/PP UAProf vocabulary that we extended with a user preferences component. Based on the UP attributes one can attach appearance conditions to the elements from CM, AM, and PM resulting in the so-called CM adaptation model, AM adaptation model and PM adaptation model. TOOL Hera Presentation Generator (HPG) is an integrated software environment that supports the presentation generation phase of the Hera methodology. It allows the graphical construction of CM, AM, and PM, the definition and instantiation of UP, and the building of a Web presentation based on Hera data transformations. EXTERNAL TOOLS HPG uses: Saxon for interpreting the data transformations specified as XSLT stylesheets. Jena for validating the Hera RDF models. Visio for implementing the CM, AM, and PM graphical builders. Figure 1. Presentation generation phase in Hera. Figure 2. Hera Presentation Generator interface. Figure 3. HPG-generated presentations for different user platforms. HPG has two main parts: AM instance generation (left-hand side of Figure 2). First CM is selected or built, and afterwards it is checked for validity. In the same way AM is selected or built, and after that it is checked for validity. After that one selects or constructs UP relevant for AM. Then, AM is unfolded (a technical step) and adapted. Based on the adapted AM, the specifications of the next transformation are generated. This transformation is used to populate the adapted AM with data given by the CM instance. The CM instance can be selected or created. Web presentation generation (right-hand side of Figure 2). First PM is selected or built, and then it is checked for validity. In the same way as for AM, PM is unfolded (a technical step) and adapted. After that one selects or constructs UP relevant for PM. Using the same technique as above, based on the adapted PM, the specifications of the next transformation are generated. This transformation is used to populate the adapted PM with data given by the AM instance. In the last transformation step the user selects the presentation format interpretable by the user browsing platform (HTML, HTML+TIME, SMIL, WML, or cHTML). Figure 3 shows an example presentation of the same data on different user platforms.