📋 Tender Overview
Natural Language Understanding For Non-Standard Languages And Dialects
Natural Language Understanding For Non-Standard Languages And Dialectswhen Artificially Intelligent Language Models And Algorithms Crunch The Numbers Of Huge Data Sets, They Are Prone To Bias Simply Because Linguistic Diversity Is Inadequately Represented. This Exclusion Involves Millions Who Converse In Dialects Or Unusual Languages. It Also Precludes Them From Emergent Future Technologies. The Eu-Funded Dialect Project Will Create Algorithms Which Facilitate High Levels Of Input Variation To Allow Diverse Dialects To Be Incorporated Into Language Technology. It Will Also Broaden Ground Truth Labels (I.E. Computer Instructions Used To Check Accuracy In The Real World) In Interactive Learning By Including Elements Of Human Uncertainty. The Result Will Be Less Data-Intensive, And It Will Make For More Equitable And Accurate Language Processing.
Funded Under
European Research Council (Erc)
Total Cost
€ 1 997 815,00
Eu Contribution
€ 1 997 815,00
Coordinated By
Ludwig-Maximilians-Universitaet Muenchen
Germany
