Diplomarbeit

Using Unstructured Data Analysis and Semantic Tagging for the Adaptation of Portal Environments

Aufgabenstellung

Today’s portals are equipped with a huge amount of content. They contain hundreds or even thousands of pages and each single page several portlets. To address this problem of information overflow, portals have to become more adaptable, adaptive and even context-ware. Such a ‘Contextual Portal’ has to take into account the context in which a user is acting and learn from his behavior in order to make assumptions about his potential behavior throughout future sessions. The aim is to let the system self-adapt to better suit the users' needs - in order to better support their typical behavior and to better manage information overflow.Today’s portals are equipped with a huge amount of content. They contain hundreds or even thousands of pages and each single page several portlets. To address this problem of information overflow, portals have to become more adaptable, adaptive and even context-ware. Such a ‘Contextual Portal’ has to take into account the context in which a user is acting and learn from his behavior in order to make assumptions about his potential behavior throughout future sessions. The aim is to let the system self-adapt to better suit the users' needs - in order to better support their typical behavior and to better manage information overflow.

To support adaptation portals need a proper contextual basis. This includes not only information about the user, his preferences, needs and device, but also about the portals and the contents they provide. Thereby, current trends referred to as “Web2.0” can be exploited, which propagate the user-driven semantic annotation, i.e. “Semantic Tagging” of content. Another approach can be the analysis of the content, which is mostly unstructured, like the user’s instant messages, emails or other documents. Open frameworks, like the Unstructured Information Management Architecture UIMA, allow for the automatic analysis and extraction of relevant semantics from these data.
Goal of this thesis is to analyze the possibilities for the user-driven semantic annotation and automatic, semantic data analysis of portals and their content in order to facilitate the adaptation of a portals navigation and page layout. In more detail, the following tasks have to be accomplished:

Analysis of the current state of the art, state of technology and related work regarding the semantic description and tagging of resources as well as the semantic analysis of unstructured data, focusing on the use in portal environments
Analysis of the requirements and possibilities for adapting a portal’s layout and/or navigation based on the semantic description of portals and its resources
Design of mechanisms for the integration of semantic tagging and unstructured data analysis tools, e.g. the UIMA framework, into portal environments.
Design of a concept for the adaptation of portal layout and navigation based on the semantics provided by both the semantic tags and UIMA.
Prototypical realization of the concept within on a real-world scenario demonstrating the benefits of the new approach.

Weitere Informationen

Autor(en):	René Pietzsch
Laufzeit:	01.04.2007 - 30.09.2007
Verantwortlicher HSL:	Prof. Dr.-Ing. Klaus Meißner
Betreuer:	Dipl.-Ing., M.Sc. Andreas Nauerz Dr.-Ing. Stefan Pietschmann
Institut, Lehrstuhl:	SMT, Multimediatechnik
Zugehörige Projekte:	Sonstiges
Zugehörige Partner:	IBM Deutschland