MEE : An Automatic Metric for Evaluation Using Embeddings for Machine Translation

We propose MEE, an approach for automatic Machine Translation (MT) evaluation which leverages the similarity between embeddings of words in candidate and reference sentences to assess translation quality. Unigrams are matched based on their surface forms, root forms and meanings which aids to captur...

Full description

Saved in:
Bibliographic Details
Published in2020 IEEE 7th International Conference on Data Science and Advanced Analytics (DSAA) pp. 292 - 299
Main Authors Mukherjee, Ananya, Ala, Hema, Shrivastava, Manish, Sharma, Dipti Misra
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.10.2020
Subjects
Automatic Metrics
Conferences
Data science
Embeddings
Frequency modulation
Morphological Languages
MT Evaluation
Semantic Evaluation
Online AccessGet full text
DOI10.1109/DSAA49011.2020.00042

Cover

More Information
Summary:We propose MEE, an approach for automatic Machine Translation (MT) evaluation which leverages the similarity between embeddings of words in candidate and reference sentences to assess translation quality. Unigrams are matched based on their surface forms, root forms and meanings which aids to capture lexical, morphological and semantic equivalence. We perform experiments for MT from English to four Indian Languages (Telugu, Marathi, Bengali and Hindi) on a robust dataset comprising simple and complex sentences with good and bad translations. Further, it is observed that the proposed metric correlates better with human judgements than the existing widely used metrics.
DOI:10.1109/DSAA49011.2020.00042