Combining User Interaction with Context Information
for Semantic Web-Site Annotation
The Semantic Shadow (SemS) concept is a model for managing contentual and structural annotations on web page elements and their values. The model supports a contextual weighting of the annotated information, allowing to specify the annotation values in relation to the evaluation context.
A procedure has been developed, which allows to manage and process these context-dependent meta information on web page elements using a dedicated programming interface. Two distinct implementations for the model have been implemented: One based on Java objects, the other using the Resource Description Framework (RDF) as modeling backend. This RDF-based storage allows to integrate the annotations of the Semantic Shadow with other information of the Semantic Web.
To demonstrate the application of the Semantic Shadow concept, a procedure to optimize web based user interfaces based on the structural semantics has been developed: Assuming a mobile client, a requested web page is dynamically adapted by a proxy prototype, whereas the context-awareness of the adaptation can be directly modeled alongside with the structural annotations.
To overcome the drawback of missing annotations for existing web pages, a concept has been developed to derive context-dependent meta-information on the web pages from their usage: From the observation of the users' interaction with a web page, certain context-dependent structural information about the concerned web page elements can be derived and stored in the annotation model of the Semantic Shadow concept.
The project source code is divided into five main fragments:
- SemS: The core fraework of the Semantic Shadow concept.
- ShadowProxy: An extendable proxy implementation coupled to the SemS framework.
- SemSAnnotator: A SWT based GUI for annotation editing.
- SemSAnalyzer: An implementation of analysis algorithms to infer annotations from web based user tracking logs.
- Op4Mo: Examples for SemS based optimizations for mobile clients.