Microsoft Translator Hub: Discussion of BLEU Score
- Chris Wendt, Microsoft; Will Lewis, Microsoft
Chris Wendt interviews Dr. Will Lewis to get a deeper insight into automatic scoring of customized machine translation systems. Microsoft Translator Hub & Bing Translator use an automatic scoring mechanism known as BLEU (Bilingual Evaluation Understudy). Dr. Will Lewis explains how this scoring mechanism works, how it compares to human evaluation and what factors impact BLEU score.
-
-
Chris Wendt
Principal Group Program Manager
-
Will Lewis
Principal PM Architect
-
-
Watch Next
-
-
-
Evaluating the Cultural Relevance of AI Models and Products: Learnings on Maternal Health ASR, Data Augmentation and User Testing Methods
- Oche Ankeli,
- Ertony Bashil,
- Dhananjay Balakrishnan
-
-
-
-
-
Making Sentence Embeddings Robust to User-Generated Content
- Lydia Nishimwe
-
MEGA: Multi-lingual Evaluation of Generative AI
- Kabir Ahuja,
- Millicent Ochieng
-
