Learning Center
About Localization

Six Ways Smartling is Leveraging Artificial Intelligence Right Now

Smartling is actively leveraging Artificial Intelligence to make your translations even better.

When we think of Artificial Intelligence (AI), it's easy to jump straight away to the AI we know from pop culture.

Who wouldn't want something akin to Iron Man's Jarvis: a life-like, yet digital, butler embedded into our gadgets that’s always ready to help us make the right decisions? While the technology hasn't hit that level of sophistication just yet, AI has already made its way into some of the tools and services we use every day.

As the pioneer of cloud translation software, Smartling is already leveraging AI to simplify and optimize your entire translation process.

Let’s take a closer look at exactly what that means.

How Do We Define AI in Translation?

We can dispel quite a bit of confusion around AI by moving forward consistently. While the semantics may differ from definition to definition, the general idea is clear:

Artificial Intelligence is an aspect of computer science that focuses on developing tools and solutions capable of performing tasks on their own.

Different forms of AI pull from different sets of data and analyze that data to uncover patterns and determine which strategy or action will have the highest probability for success.

For example, we often think of machine translation at first. The most modern machine translation models are known as Neural Machine Translation.

machine-translation These engines directly leverage deep machine learning neural networks to determine how to properly translate your content by interpreting the intent of that source content. The result is an AI that acts more in-line with how a human translator works, rather than just a bilingual dictionary.

Artificial Intelligence Lives In Your TMS

While NMT is certainly impressive, AI in cloud translation software focuses specifically on translation management. This means that the AI is tied directly to the concept of automated translation, and is leveraged within specific instances to improve both translation efficiency and quality.

Automated translation and AI aren’t about removing the need for human translators, but rather supporting them, and simplifying the process from A to Z.

More specifically, AI is used to help you work smarter, not harder, while improving translation quality throughout the process. This means content can be pushed to market at a much more rapid pace without sacrificing quality.

1. Machine Translation
NMT represents the most widely known use of AI in translation. We already see powerful engines from machine technology providers like Google, Microsoft, and Amazon -- especially Google's latest app.

2. Natural Language Processing
Generally, the tech that works alongside an NMT engine, Natural Language Processing is all about converting human speech into a computer's binary language. What's important here is that NLP can be used to scan and analyze documents to uncover any potential errors, improvements, or even document classification on a mass scale.

3. Process Automation
When it comes to real-world, usable AI, process automation is going to be where we find the most obvious use-cases. A more traditional translation process, and the related workflows, are slow and error-prone. Instead, automated translation relies on AI to handle more basic and repetitive tasks assigning jobs, reviewing and revising content, submitting content, or rejecting content altogether.

AI in Smartling

The best part about artificial intelligence in translation, besides being apart of the futuristic world we live in, is the overt opportunity to reduce cost while saving valuable time -- a major benefit behind cloud translation software.

For example, with the ability to preconfigure job automation rules, users can keep content free-flowing, relying on automated quality checks to identify any issues within the translation.

using-smartling-tile

According to the latest CSA MarketFlex Report for Language-Oriented TMS, Smartling is constantly leveraging "linguistic and project data" to teach our services and tools how to work even better.

With AI constantly involved, users can optimize the efficiency of their translation process while learning how to improve even further. Here are six real-world features, illustrating how Smartling leverages Artificial Intelligence in translation to constantly improve both time-to-market and translation quality:

1. Automated project management

Smartling surfaces relevant data, and makes automated decisions about how to best translate content from one language to any target language.

Project management can be automated based on a set of pre-programmed parameters, including budget, time-to-market, and quality expectations.

  • Job Creation - Schedule content to automatically be packaged into a job, based on a set of rules including file type, project type, languages and more.
  • Content routing - Automatically route and deliver content to the translation supplier of your choice, after the content has been assigned as a job.
  • Notifications - Smartling notifies the next user when work is ready for translation, editing or review so nothing slips through the cracks.

2. Quality Checks

Users can configure their own unique automated processes and task management through customizable Quality Checks.

Instead of always requiring human review, users can determine the priority of errors, and require action to be taken to resolve these errors before content can move to the next step.

Hence the term automated translation. Instead of requiring a constant human editing process, AI can step in to help clean up content and remind translators of any errors during the translation process itself.

3. Dynamic workflows

With dynamic workflows, users can design their own workflows based on the unique needs based on their role during the translation and content management process. Users can even link multiple actions to one trigger.

For example, an advanced Dynamic Workflow can determine when a human translator might be a better fit for specific content.

One method available is to leverage an existing Translation Memory (TM) to determine the complexity of the content. When Smartling compares source content against your brand's TM, it will often find word matches that are not 100% identical. These are known as a Fuzzy Match, and are represented as a percentage based on how closely the words align in meaning.

Content with a high number of Fuzzy Matches that score below 90% can automatically be routed to a translator instead of a Machine Translation workflow. The idea here is that Smartling can help determine the best path to take for success based on your content and translation history.

The goal is to achieve accurate and high quality translation; Dynamic Workflows enables you to do this without manually managing every string.

4. Data and Analytics

Our machine learning algorithm compiles an automated quality score for every translation as it moves through Smartling.

The result: a real-time view of translation quality across all your projects, languages and translation providers.

Dynamic and Velocity Workflow Reports visually represent the potential benefit these workflows added to the translation process, or where greater optimization is required.

5. Smartling Draft

One of our favorite AI additions to Smartling has to be Draft. Just recently released, Draft represents a major push in utilizing AI.

Draft is meant to unify the entire global content creation experience. Draft, an automated writing collaboration tool, will actively scan your content as it is being written, and offer suggestions on how to improve your source content for the highest quality translation possible.

The best part is that it all happens in real-time by pulling from available data within your translation memory to suggest phrases or words that best fit within the content.

6. Quality Confidence Score

Smartling's Quality Confidence Score (QCS) provides a prediction of translation quality results. Based on a score of 1-100, the QCS helps users recognize how close or far away the quality of their piece is from "professional quality," meaning that if content were to be manually evaluated by a human agent.

This score tracks multiple metrics, with a granular view of each, including Translation Memory leverage, Glossary terms, Workflow Steps, Quality Check Errors, String Length, Visual Context, and any other issues within the content itself.

The Smartling Advantage

Artificial intelligence in translation is a bit of a hot topic at the moment, following such a boom in real-world use of the technology. It is always important to manage expectations around emerging technologies.

This holds true when we discuss how AI is impacting both the translation process, as well as the localization industry as a whole.

When it comes to Smartling, AI is apart of almost everything the service does. Smartling is constantly working to improve your translation workflow process while learning how to improve the service itself at the same time.

If you want to learn more, give us a call, we’re more than happy to help you move the world with words.

About Matt

Matt Grech is the Content Marketing Manager at Smartling, responsible for growing Smartling awareness and brand content. As a digital content writer, Matt applies his journalistic lens to content, helping users deepen their understanding of the brand, services and technology provided by Smartling. Matt has previously contributed to an industry leading Unified Communications resource, as well as local newspapers where he developed his unique ability to investigate, interview, and transform complex problems into simple solutions.

View our help center

Read our docs

Read our blog