Unbabel, the AI-powered, human-refined translation platform that enables multilingual customer service at scale, announced the release of COMET (Crosslingual Optimized Metric for Evaluation of Translation), an open-source neural framework and metric for Machine Translation (MT) evaluation that has been validated as a top performing metric by the 2020 Fifth Conference on Machine Translation (WMT20).
COMET reduces the need for human review, enabling rapid assessment and deployment of accurate machine translation models for the benefit of Unbabel’s customer service customers.
Existing MT evaluation solutions correlate poorly with human judgements of translation quality. An MT translation might misfire with regards to syntax, grammar or other important linguistic elements, leading to miscommunication, a poor brand impression by customers and, in the worst case, offensive communication.
COMET, the successor to Unbabel’s Alon Lavie’s innovative MT evaluation metric METEOR (Metric for Evaluation of Translation With Explicit ORdering), stands to replace both METEOR and BLEU (Bilingual Evaluation Understudy) as the modern metric for measuring MT quality. COMET captures the meaning similarity between texts with enough granularity to accurately predict human experts’ translation quality judgments. It takes advantage of recent breakthroughs in large-scale cross-lingual neural language modeling, resulting in multilingual and adaptable MT evaluation models of unprecedented accuracy.
“We are launching COMET as an open-source, ready-to-use, trained model because it can greatly help drive and accelerate MT research and development to levels of accuracy not seen before. We believe that COMET should be adopted as a new standard measure for assessing the quality of MT systems across multiple languages,” said Alon Lavie, vice president of language technologies at Unbabel, co-creator of METEOR and consulting professor at Carnegie Mellon University. “Unbabel is deeply committed to maintaining its leadership in this space and removing the misconception that MT means low quality when it comes to translation.”
Unbabel processes high volumes of translations using highly specialized AI models for customer service solutions. Applied in a wide range of customer domains, the company’s MT engines are continually retrained to ensure that the highest levels of translation quality and robustness.
COMET enables developers and implementers of MT to measure the quality of machine-generated translations in a more accurate and standardized way for 100+ languages. The framework will allow Unbabel to further automate its MT evaluation processes and deliver ever-improving MT to the customer service industry.
COMET’s open source project is hosted in GitHub: https://github.com/Unbabel/COMET.
To learn more about COMET, please visit: https://resources.unbabel.com/comet-unbabel-research.
Unbabel removes language barriers by blending advanced artificial intelligence with real time, human translations. The company’s enterprise translation platform enables customer-centric brands to provide multilingual customer service and experience at scale – delivering native-quality translations quickly, efficiently and smarter over time. It transforms customer contact centres from a cost to profit centre, providing automated translation, quality estimation, human expertise and CRM integrations. With the vision to become the World’s Translation Layer, Unbabel works with leading brands such as Facebook, Microsoft, Booking.com and Uber, to make their customers happier and their support operations more efficient.
For more information visit www.unbabel.com.