Generating resource profiles by exploiting the context of social annotations

Fabian Abel presenting @ISWC2011

Typical tagging systems merely capture that part of the tagging interactions that enrich the semantics of tag assignments according to the system’s purposes. The common practice is to build tag-based resource or user profiles on the basis of statistics about tags, disregarding the additional evidence that pertain to the resource, the user or the tag assignment itself. Thus, the main bulk of this valuable information is ignored when generating user or resource profiles.

In this work, we formalize the notion of tag-based and context-based resource profiles and introduce a generic strategy for building such profiles by incorporating available context information from all parts involved in a tag assignment. Our method takes into account not only the contextual information attached to the tag, the user and the resource, but also the metadata attached to the tag assignment itself. We demonstrate and evaluate our approach on two different social tagging systems and analyze the impact of several context-based resource modeling strategies within the scope of tag recommendations. The outcomes of our study suggest a significant improvement over other methods typically employed for this task.

Venue: ISWC2011

Presentation: video and slides (by Fabian Abel)

Authors: Ricardo Kawase, George Papadakis, Fabian Abel

PDF: kawase.iswc2011

Classification of user interest patterns using a virtual folksonomy

User interest in topics and resources is known to be recurrent and to follow specific patterns, depending on the type of topic or resource. Traditional methods for predicting reoccurring patterns are based on ranking and associative models. In this paper we identify several ‘canonical’ patterns by clustering keywords related to visited resources, making use of a large repository of Web usage data. The keywords are derived from a ‘virtual’ folksonomy of tags assigned to these resources using a collaborative bookmarking system.

Venue: JCDL2011

Authors: Ricardo Kawase and Eelco Herder

PDF: kawase-jcdl2011a

Leveraging Search and Content Exploration by Exploiting Context in Folksonomy Systems

With the advent of Web 2.0 tagging became a popular feature in social media systems. People tag diverse kinds of content, e.g. products at Amazon, music at Last.fm, images at Flickr, etc. In the last years several researchers analyzed the impact of tags on information retrieval. Most works focused on tags only and ignored context information. In this article we present context-aware approaches for learning semantics and improve personalized information retrieval in tagging systems. We investigate how explorative search, initialized by clicking on tags, can be enhanced with automatically produced context information so that search results better fit to the actual information needs of the users. We introduce the SocialHITS algorithm and present an experiment where we compare different algorithms for ranking users, tags, and resources in a contextualized way. We showcase our approaches in the domain of images and present the TagMe! system that enables users to explore and tag Flickr pictures. In TagMe! we further demonstrate how advanced context information can easily be generated: TagMe! allows users to attach tag assignments to a specific area within an image and to categorize tag assignments. In our corresponding evaluation we show that those additional facets of tag assignments gain valuable semantics, which can be applied to improve existing search and ranking algorithms significantly.

Venue: The New Review of Hypermedia and Multimedia 2010

Authors: Fabian Abel, Matteo Baldoni, Cristina Baroglio, Nicola Henze, Ricardo Kawase, Daniel Krause and Viviana Patti

PDF: abel-nrhm2010

Leveraging multi-faceted tagging to improve search in folksonomy systems

In this paper we present ranking algorithms for folksonomy systems that exploit additional contextual information attached to tag assignments available. We evaluate the algorithms in the TagMe! system, a tagging front-end for Flickr, and show that our algorithms, which exploit categories, spatial information, and URIs describing the semantics of tag assignments, perform significantly better than the FolkRank that does not consider such contextual information.

Venue: HT2010

Authors: Fabian Abel, Ricardo Kawase and Daniel Krause

PDF: abel-ht2010

The Art of Multi-faceted Tagging – Interweaving Spatial Annotations, Categories, Meaningful URIs and Tags

In this paper we present TagMe!, a tagging and exploration front-end for Flickr images, which enables users to attach tag assignments to a specific area within an image and to categorize tag assignments. We analyze the differences between tags and categories and show how both facets can be applied to learn semantic relations between concepts referenced by tags and categories. TagMe! automatically maps tags and categories to DBpedia URIs to clearly define the meaning. In our experiments we compare different strategies to realize such semantic mappings and show that already lightweight approaches map tags and categories with high precisions (86.85% and 93.77% respectively). We further discuss how multi-faceted tagging helps to improve the retrieval of folksonomy entities. The TagMe! system is currently available at http://tagme.groupme.org

 

Venue: WEBIST2010

Authors:  Fabian Abel, Ricardo Kawase, Daniel Krause, Patrick Siehndel and Nicole Ullmann