Collaborative tagging mechanisms are integral to social computing applications in a variety of domains. Their expected benefits include simplified retrieval of digital content, as well as enhanced ability of a community to makes sense of the shared content. We examine the impact of collaborative tagging in context of nutrition management. In a controlled experiment we asked individuals to assess the nutritional value of meals based on photographic images and observed the impact of different types of tags and tagging mechanisms on individuals nutritional sensemaking. The results of the study show that tags enhance individuals' ability to remember the viewed meals. However, we found that some types of tags can be detrimental to sensemaking, rather than supporting it. These findings stress the importance of tagging vocabularies and suggest a need for expert moderation of community sensemaking.

We present contextual facets, a novel user interface technique for navigating websites that publish large collections of semi-structured data. Contextual facets extend traditional faceted navigation techniques by transforming webpage elements into user interface components for filtering and retrieving related webpages. To investigate users' reactions to contextual facets, we built FacetPatch, a web browser that automatically generates contextual facet interfaces. As the user browses the web, FacetPatch automatically extracts semi-structured data from collections of webpages and overlays contextual facets on top of the current page. Participants in an exploratory user evaluation of FacetPatch were enthusiastic about contextual facets and often preferred them to an existing, familiar faceted navigation interface. We discuss how we improved the design of contextual facets and FacetPatch based on the results of this study.

Social network-based systems usually suffer from two major limitations: they tend to rely on a single data source (e.g. email traffic), and the form of network patterns is often privileged over their content. To go beyond these limitations we describe a system we developed to visualize and navigate hybrid networks constructed from multiple data sources - with a direct link between formal representations and the raw content. We illustrate the benefits of our approach by analyzing patterns of collaboration in a large Open Source project, using hybrid networks to uncover important roles that would otherwise have been missed.