Harvesting Knowledge from Social Annotations


Kristina Lerman
Thursday, October 13, 2011
4:15 PM – 5:30 PM
Rose Hills Theater
More information
Pomona colloquium information

Social Annotation captures the collective knowledge of thousands of Social Web users and can potentially be used to enhance an array of applications including search, personalization and recommendation. In order to make best use of social annotation we need methods that effectively deal with the challenges of data sparseness and noise, as well as take into account inconsistency in the vocabulary, interests, and the level of expertise among individual users. We present computational approaches to extracting knowledge from structured annotations created by thousands of users of the social photo-sharing site Flickr. First, I present the folksonomy learning problem, i.e., learning a common taxonomy from many shallow personal hierarchies created by individual Flickr users. I describe a novel probabilistic approach, based on affinity propagation, that allows us to integrate structural constraints into the inference process in order to combine many smaller structures simultaneously into a larger common structure while avoiding structural inconsistencies. Second, I present some recent results of exploiting geo-referenced metadata on Flickr to learn places and relations between them.

Kristina Lerman is a Project Leader at the Information Sciences Institute and holds a joint appointment as a Research Assistant Professor in the USC Viterbi School of Engineering’s Computer Science Department. Her research focuses on applying network- and machine learning-based methods to problems in social computing.