Generating and Optimizing Graphical User Interfaces for Semantic

Generating and Optimizing Graphical User Interfaces for Semantic Service Compositions Eran Toch, Iris Reinhartz-Berger, Avigdor Gal and Dov Dori Motivation Dynamic compositions of semantic Web services result in a challenge for user interface engineering: • Service composition is driven by semantic precision • User Interface is driven by usability Existing approaches [Khushra & Lassila, Puerta & Eisenstein] provided a model of ontology-based generation of user interfaces. Creating usable user interfaces for dynamic compositions is still a challenge. 1. A semantic Web composition is created by a search engine Get FOAF Address Get FOAF Profile Google Map API output input 2. The ontology and process model of the composition are used in order to create a Webbased prototype input FOAF Profile Name URL Semantic Web Services output Map Location FOAF (friend of a friend) Ontology Lat/Long Ontology Geography Ontology 3. Users can navigate through the prototype, which supports automatic validation and online help which are based on the ontology structure [Khushra & Lassila] Deepali Khushra j and Ora Lassila. Ontological approach to generating personalized user interfaces for web services. In International Semantic Web Conference, pages 916– 927, 2005. [Puerta &Eisenstein] A. R. Puerta and J. Eisenstein. Towards a general computational framework for model-based interface development systems. Knowledge-Based Systems, 12: 433– 442, 1999. [Yahoo] Yahoo Design Patterns, http: //developer. yahoo. com/ypatterns/ [Borchers] Jan Borchers. A Pattern Approach to Interaction Design. John Wiley & Sons, Inc. , 2001. Foreword By-Frank Buschmann. Optimizing Usability As semantic Web service compositions are not constructed with usability in mind, they might be hard to use. For example: Liquid-Interface uses formal models of UI design patterns [Yahoo, Borchers], in order to optimize the overall usability of the prototype’s navigation structure. Naïve. Task Model Example: Flat and Narrow Tree design pattern P 1 P 2. . . Pn opt Optimized Task Model The optimized task model exhibit the maximal usability score: 1 page with 25 input fields The graph represents the usability score of the pattern, as a function of the depth of the navigating the application. 3 is the optimal depth in this example. T – task model Pi(T) – usability score of pattern i on task model T wi – weight of pattern i 5 pages with 7 fields