Curated Machine Translation (MT) initially drives value through the improvement of voice, tone, and style of the MT output. United Language Group uses the latest developments in Neural Machine Translation (NMT) and Natural Language Processing (NLP) to customize your machine translation engines to your business verticals. This enables faster turnaround times and cost efficiencies within the overall translation cycle.
United Language Group approaches machine translation in a unique way. We start with testing to see if the outcomes have achieved your intended purpose. Based on the testing score, we determine if machine language translation services are right for your specific vertical, content type, and language. Alternatively, the test may indicate where a machine translation software solution may require additional training to become proficient. Ultimately, our assessment determines where your organization should and should not utilize machine translation based on your expectations and established parameters. After our evaluation, we help to deploy and integrate the best machine translation services into your workflow, but we don’t stop there. We continue to train and test your machine translation engine to improve quality and set up a continuous improvement loop to monitor performance.
United Language Group’s advanced machine translation testing model is a leader in the industry. It is designed to provide critical data to our data scientists and enable you to effectively verify the value your organization could receive from the utilization of a curated Machine Translation Solution.
Our testing model includes:
Automated testing with BLEU (BiLingual Evaluation Understudy) score and Edit Distance.
Human linguist feedback and evaluation scores.
Machine Translation / Translation Memory (TM) Concordance to show the percentage of the changes to segments.
Analysis regarding editing times required to get the content to full human quality.
Blind evaluation of human edit scores and preferential changes against human-provided content (helps effective weighting of review feedback).
Our Raw MT solution is best for when you need real-time translation, such as live chats or high-volume content that is time-sensitive. With Raw MT, we focus on training beforehand because the content does not undergo human review.
Our machine translation post editing services solution is best used when seeking accuracy, but with a focus on improving turnaround times and driving cost efficiency. A human review or a post edit (PE) is performed on MT PE to confirm there are no problematic translation errors.
Our MT TE solution is best in scenarios where the content is highly-sensitive, but you are still wanting to drive cost efficiencies. MT TE goes through two human reviews because the content requires two quality checkpoints and needs to reach full human quality.
At United Language Group we combine best-in-class technology with optimized workflows and advanced automation to provide you with efficient, accurate and cost-effective translations of your medical claims.