Interactive IR amplifies the issues related to learning from interaction with the data, trust, and reconciliation of multiple perspectives on relevance issues. Recently we conducted experiments that involve relevance assessment of large documents (books) through extensive interaction with the content, search engine, and multiple human assessors. This exercise made us revisit the notion of relevance, trust, and authoritativeness in relevance judgements. I suggest that we discuss this theme as it touches on the very foundations of interactivity: evolving view of the topic scope and relevance assessment.   

I would also like us to give some attention to discussing interactive retrieval as part of the content management workflows. For example, lots of information access is aimed at revisiting the content that has been used in the recent or more distant past (desktop data, mail, online blogs and community data, web, etc.). Some content revisitation is better supported by browsing while other requires search. In order to support efficient access to content we need to look at the practices around interaction with the content during complex tasks that users perform (not restricted to search and browsing). We may be able to capture additional context by leveraging user’s need to manage their work efficiently and thus provide more metadata about their activities. We can then use the data for better data revisitation. I can share results of our recent work on introducing tagging into the desktop environment and showing how new ways of browsing and search are enabled.