Natural language research announcements

["Alexandre Klementiev" <aklement AT gmail.com>]

Title:  Unsupervised Rank Aggregation with Distance-Based Models


Speaker:  Alexandre Klementiev, UIUC






Date: 
4/22/08, 4-5pm
Location:      3403 Siebel Center

Abstract:
Consider the scenario where each member of a panel of judges
independently generates a (partial) ranking over a set of items while
attempting to reproduce a true underlying ranking according to their
level of expertise. This setting motivates a fundamental machine
learning and information retrieval problem - the necessity to
meaningfully aggregate preference rankings into a joint ranking. For
example, in meta-search the aim is to aggregate Web search query
results from several engines into a more accurate ranking. In many
natural language processing applications, such as machine translation,
there has been an increased interest in combining the results of
multiple systems built on different principles in an effort to improve
performance.

Although a number of heuristic and supervised learning approaches to
rank aggregation exist, they require domain knowledge or supervised
ranked data both of which are expensive to acquire. We propose a
mathematical framework for learning to aggregate (partial) rankings
without supervision. We instantiate the framework for the cases of
combining permutations and combining top-k lists, and propose a novel
metric for the latter. Experiments in both scenarios demonstrate the
effectiveness of the proposed formalism.