Adam Muzika, sales manager at SDL and focused on the legal industry, covered some of the key eDiscovery trends and challenges facing legal firms. As Adam points out, “An eye-watering 73% of e-discovery costs are attributed to document review alone.” His blog looked at how the latest advances in Artificial Intelligence (AI) and Machine Translation are changing the game for the industry, presenting a golden moment to combined technologies in a way that takes away the sheer time and cost involved in electronic discovery for legal proceedings.
Unsurprisingly, Machine Translation proved to be a popular topic in 2019. SDL’s language technology evangelist, Kirti Vashee, is a fount of knowledge on this subject, and tackled the open source MT debate by highlighting the benefits—and restrictions—of going it alone and building your own MT model, or adopting a robust enterprise MT platform. “The evidence suggests that embarking on a self-managed open-source-based MT initiative is for the very few who are ready to make the substantial long-term commitment”, explains Kirti. However, even investment in private MT engines are not risk free—security, interoperability and the quality of data needs to be considered.
Let’s face it, companies just can't keep up with their content demands. Especially those trying to engage with consumers across any one of their 3 connected devices, who spend on average 6.5 hours a day online. SDL’s VP of product marketing, Marcus Hearne, introduces the concept—and reality—of the intelligent translation era, where companies can adopt a machine-first, human-optimized approach to transform their globalization strategies. Forget going global, it’s about going ‘omni-market.’ As Marcus explains, “rather than localizing top-tier content exclusively for established markets, translate everything for everyone and see where the chips fall. Then let humans take over to determine how content should be tailored to resonate further with the audience.”
Kirti Vashee looks at another hot topic impacting the world of Machine Translation: quality. Most businesses pick an MT platform based on two criteria—cost, and best quality. Cost is clear, but as a brand how can you assess the quality of a potential platform’s MT output? While BLEU and Lepor scoring techniques can provide a useful guidance, Kirti suggests that companies should go beyond such simple assessment and factor in non-linguistic attributes, including adaptability, manageability, security, integration and deployment flexibility. “It is more useful to focus testing efforts on specific enterprise use case requirements," explains Kirti as he outlines a more meaningful evaluation framework.
To translate, or transcreate—that’s the question facing many localization teams. In this blog by Izabella Iizuka, a spectrum of translation methods are offered to help guide anyone wondering whether translation or transcreation is right for them. If you just need to understand sentiment, of perhaps online discussions and forums, then Machine Translation is perfectly suited. If you’re translating adverts or high-value content—then that human touch will be needed. But where do you draw the line? And do you always need to translate your content? You might be surprised to learn that sometimes leaving the content in the original language might actually be better than translating—or transcreating.
At SDL’s flagship customer event, SDL Connect, 400 delegates gathered over the course of two days to learn more about the latest developments in the content and language industry. Maria Hudson, VP of corporate marketing and communications, provided an overview of the show, wrapping up some of the key topics, events and presentations that took place. The blog recaps some of the big product announcements, and how they fit into SDL’s strategy of building true intelligence into our content and language solutions. If you missed this year’s event this is a great read if you want to find out what happened, and why you should attend next year’s event.
If you’re a translator or project manager, you’ll no doubt know how quickly things can spiral out of control when dealing with multiple translation projects, tight deadlines and dozens of stakeholders. Andrew Thomas, senior director of product marketing, reveals the much-anticipated launch SDL Language Cloud, designed to accelerate and automate translation for all content types, languages and translation methods. Andrew explains in his blog how SDL Language Cloud is the first cloud-based solution enabling companies to benefit from both machine intelligence and human expertise, optimizing the translation process for greater control, visibility and scalability across the global content supply chain. And by unifying SDL’s extensive language services, robust translation management capabilities and cutting-edge Neural Machine Translation, SDL Language Cloud can support any business with their global content translation demands.
And topping this year’s list of most read blogs is by Kirti Vashee, looking in detail at the BLEU scoring methodology, a way of measuring the quality of Machine Translated content. The piece explains the practicalities—and limitations—of the methodology, alongside how it should be used when evaluating Machine Translated content. Kirti acknowledges some of the criticisms from the MT community, alongside his thoughts on what the future holds for the method. It seems that, only through human validation—combined with BLEU (and indeed other upcoming methodologies), can companies gain a true and realistic picture of their MT content.
Clearly, MT is the most popular subject across the year. But what we found interesting, is the curiosity among readers for insight on how to best evaluate MT content. And particularly, how MT is opening real opportunities to improve eDiscovery within the legal industry.
A big thank you to all our readers for following us throughout the year!