IR systems in many cases implement an ad-hoc combination of modalities, features, and interpretation of object semantics with interpretations of user needs and queries. In this talk I want to raise questions concerning a systematic aproach as to

\begin{itemize}
  \item where the information resides in a digital object and its environment, i.e. which aspects$/$dimensions$/$modalities of information here are in any specific setting,

  \item why it exists and is being searched for, i.e. for which purpose it was created by whom, which context it is living in, and what the intention of a searcher might be, and

  \item ultimately, what information actually is, which incarnations of representations it exists in, how it can be indexed and described in various modalities even if the task at hand may seem inherently single-modal
\end{itemize}

I will use scenaios and show examples from three different domains to illustrate the above questions, namely


\begin {itemize}

  \item \bf{Music Retrieval}, with a focus on the multi-modal aspects of music information, including textual, image and video dimensions, and potential combinations of those to capture the complexities of seemingly mono-modal (acoustic) information in real-world settings

  \item \bf{Web Archiving}, with a focus on ethical aspects of Web Arcive access and consequences of$/$for search functionalities and machine learnign algorithms analyzing Web Archive content

 \item \bef{Digital Preservation}, focussing on what actually is the information represented by and in an object that is to be preserved (and thus the target of retrieval) in terms of significant properties, context of information and authenticity, going beyond the simple mono$/$multi-modal search of primary content


\end{itemize}

These example demonstrate numerous issues calling for a consolidated view of information and he information retrieval process. They call for an integrated model for matching information representation, structuring and interpretation across a range of orthogonal dimensions, with use cases detailing user needs and intentions. Only by providing such an integrated view and structured matching we may be able to achieve a consolidated means of benchmarking and evaluating interactive information retrieval for the complex tasks encountered in real-world settings.