Material Detail

Multilingual Document Retrieval Through Hub Languages

Multilingual Document Retrieval Through Hub Languages

This video was recorded at 15th International Multiconference on Information Society, Ljubljana 2012. In this paper we extend previous work on document retrieval across multilingual corpora. In this setting it is often assumed that we have a certain alignment given based on which we can learn mapping between spaces. In true multilingual corpora however, we often do not have alignments between all languages. There are hub languages which have alignments with many other languages. We look at the effectiveness of leveraging these alignments to learn maps which may have small or no alignments given. We test several methods and investigate the performance of various approaches on theWikipedia dataset.


  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material


Log in to participate in the discussions or sign up if you are not already a MERLOT member.