Application Framework with Demand-Driven Mashup for Selective Browsing

We are developing a new mashup framework for creating flexible applications in which users can selectively browse through mashup items. The framework provides GUI components called widgets through which users can browse mashed-up data selectively, and the system processes demand-driven creation of mashed-up data upon receiving access requests through widgets. The application developer has to only prepare a configuration file that specifies how to combine web services and how to display mashed-up data. This paper proposes a revised widget model for effective data display, and introduces practical applications that allow selective browsing. The revision of the widget model is to accept various GUI components, process user interactions, and provide cooperative widgets. To avoid conflict with lazy data creation, we introduce properties into widgets that are automatically maintained by the system and can be monitored by other widgets. The case study through the applications shows the situations where the initially browsed data helps users to terminate redundant searches, set effective filter settings, or change the importance of the criteria. Some applications display synoptic information through columns, maps, or distribution charts; such information is useful for selective browsing.