The Information Retrieval and Machine Learning (IRML) Group's main research goal is to contribute to the development of a statistical and information-theoretical foundation for today's information technologies and information infrasturcture.

The IRML Group develops enabeling technologies for analysing, structuring, organizing, and visualizing large content repositories, including hypertext and multimedia databases. These technologies are utilized to design tools to automatically annotate, classify, filter, retrieve, and deliver content. Special emphasis is put on personalized information access as well as robust and efficient methods of human-computer interaction.

Machine Learning methods are of particular relevance in this context, since they are crucial for developing innovative methods and tools to support intelligent forms of information access.

More specifically, current projects deal with the following topics.

• Statistical unsupervised learning methods for matrix decomposition, dimension reduction, and clustering.
• Advanced techniques and tools for intelligent query-based information retrieval that take semantic relatioships between terms into account.
• Personalized retrieval and collaborative filtering techniques that combine content analysis with models of user interests.
• Multimodal and multimedia information retrieval with special emphasis on combining text and image data.
• Methods for categorizing document and for organizing content into taxonomies.
• Visualization technologies to support interactive navigation in information spaces.
• Semantic models of hyper-lnked information repositories, like the World Wide Web, including methods for focused Web-crawling and to find Web communities.
• Enabeling technologies for distributed information agent systems.
• Retrieval of spoken documents.