| Summary: | 博士 === 國立暨南國際大學 === 資訊工程學系 === 93 === Authoring interactive hypermedia presentations with good aesthetic quality and affective quality is considered as an interdisciplinary task that usually includes intensive collaborations between designers and computer engineers. An adaptable system that changes the hypermedia presentation style automatically based on a user’s emotion request should be able to alleviate the complexity of such hypermedia authoring process. This thesis explores and combines various research perspectives, including hypermedia technology, affective computing, Kansei Engineering, web technology, interface design and artificial intelligence, so as to develop an “adaptable styling system for affective hypermedia presentations”.
There are three major parts of the thesis. The first part of the thesis describes the development of an adaptable hypermedia presentation authoring system. Separation of structure and presentation can effectively be realized by using XML, in combination with standardized style sheet languages. The system realizes the “single source, multiple delivery publishing model” in the current Web environment. Two major disadvantages with this approach include: (1) the lack of extensibility for the logical structure of the exhibition element; (2) the difficulties for authoring XSL documents.
In Part II, we design a modularization framework of hypermedia presentations. Innovative aspects of this framework include document logical structure model and presentation structure model based on Fine-Grained Styling Modules (FGSM). The approach proposes a mechanism to decompose a complicate hypermedia presentation into a number of “building blocks” that can be integrated with the ease to generate aesthetic hypermedia presentations. Based on the proposed approach, a sophisticated structured authoring system, HyPMod System (Hypermedia Presentation Modularization system) has been conducted and certified according to the performance of such system. By greatly reducing the number of hypermedia authoring operations, the human authoring process is now possible to be simulated by using a computer system. The realization of the system indeed helps us march towards an adaptive styling system for affective hypermedia presentations.
Based on the HyPMod System, in the third part, we then move forward to the design of an intelligent hypermedia styling system, Aff-HyPMod System (Intelligent Emotion Hypermedia Presentation Modularization System), which is based on Kansei engineering approach, neural network, and genetic algorithm. Innovative aspects of the system include an intelligent background image generator, color combination scheme based on affective request, the geometric fitness verification mechanism among styling modules, content, and page layout. The integration of feedforward neural network and genetic algorithm provides the mechanism for searching for the reasonable combination patterns of the styling modules that might result in a desired emotion status. The Aff-HyPMod system is the first ever reported sophisticated intelligent styling system for affective hypermedia presentations.
|
|---|
