We live in a data-centric world. We’re now creating, capturing, copying and consuming data at an unprecedented rate, reaching an enormous 79 zettabytes of data globally in 2021. For context, 1 zettabyte equals 1 sextillion bytes. The text you are reading now is less than 10 kilobytes. The Bible is approximately 4 megabytes which means that humanity has generated 20 trillion Bibles in terms of the size of text. But what does this mean for you? How can you make all this data work to your advantage? And how can AI and NLP be harnessed to solve the problems businesses face today?
What is NLP AI, and why is it important?
Let’s begin by defining exactly what we mean by AI and NLP. As Oracle puts it, Artificial Intelligence “refers to systems or machines that mimic human intelligence to perform tasks and can iteratively improve themselves based on the information they collect.” Its aim is to replicate the way humans think and to carry out specific tasks in order to free up humans to focus on other activities. A key attribute of AI is that it ‘learns,’ getting better and better as it’s trained on more and more data. AI is used in myriad different ways to streamline day-to-day business operations, from resolving queries faster with chatbots to drawing conclusions from customer reviews more easily using opinion mining. Aware of AI’s potential, businesses are investing heavily in this area.
One form of AI that’s of particular interest to the language services industry is natural language processing (NLP). According to SAS, NLP “is a branch of artificial intelligence that helps computers understand, interpret and manipulate human language.” In other words, computers analyze human language, using statistical data to draw conclusions and mimic the way humans write and speak. As we’ll see further down, NLP is used in a variety of tools throughout the localization process to save time and money.
A key attribute of AI is that it ‘learns’ and improves continuously as it’s trained on more and more data. The value of AI comes in how it can be used in a myriad of different ways to streamline day-to-day business operations, from resolving queries faster with chatbots to drawing conclusions from customer reviews more easily using opinion mining. Aware of the potential of AI, businesses are investing heavily in this area. In fact, Gartner forecasts that worldwide AI software revenue will total $62.5 billion in 2022, an increase of 21.3% from 2021.
What problems does NLP AI solve?
The biggest challenge facing the language services industry today is the need to deliver more content, faster. NLP AI enables us to make use of the vast quantities of data we’ve been accumulating to do just that. It powers a plethora of localization tools to help us deliver more content without increasing spend or staff numbers.
Machine translation is the most notable application of NLP AI in localization. Today’s NLP technology feeds into the latest generation of machine translation engines such as SYSTRAN and Intento to dramatically improve machine translation output, resulting in decreased human translation time and spend. And because modern machine translation saves so much time, organizations are better able to meet the ever-increasing demands for localized content, and in turn, better address customers’ needs.
Machine translation is what’s known as an end-to-end solution, meaning it’s fully automated and requires no user interaction. The other main approach in AI is analysis-based. In the analysis approach, NLP forms the basis of tools to enhance human translation. For instance, NLP is used to assess the quality of completed translations, flagging potential translation errors and rating overall translation quality. This helps save time and reduce the risk of human error.
What is XTM’s bespoke AI?
So we now know how NLP can help boost localization productivity in general terms. But how does it work exactly? To help you understand, I’ll explain the approach we’ve taken here at XTM.
XTM makes the most of both AI approaches: providing its users with end-to-end solutions, such as machine translation; and making the work of the human translator easier and more efficient by providing specific tools underpinned by linguistic analysis.
In 2020, we launched a new NLP AI-based technology, Inter-Language Vector Space (ILVS). Based on extensive work by Google and Facebook and using the entire internet as a dataset, ILVS calculates the probability of a target language word being the correct translation of a source language word. ILVS was revolutionary. It allowed us to introduce new features to aid the human translation process.
But we didn’t stop there. To meet emerging business demands, in 2021 we launched Multi-Faceted Translation Analysis (MFTA). The idea behind MFTA was to make word alignment more accurate. But it’s since become the framework on which all of XTM’s AI-powered tools are built. It uses different analysis tools to generate different types of data and is designed to hold any type of linguistic information and convert it into usable formats.
Watch MFTA in action:
MFTA drives a number of practical features within XTM. For a start, it can help you detect potential translation errors. And then there’s the auto-inline feature, which automatically adds tags and formatting where appropriate, allowing translators to focus on the translation itself. Meanwhile, the auto-aligner tool creates a translation memory from legacy documents which can be used in future translations, helping to save time and ensure consistency. During the alignment process, the tool also extracts terminology, creating a specialized glossary with industry-specific terms and their target language equivalents. This feature is currently available for 50 different languages.
How can your business benefit from XTM’s bespoke AI?
We’ve talked about XTM’s AI-powered features and functionalities. Now let’s look at what all this means in terms of genuine business benefits.
- Improved translation quality: Because AI takes care of mundane repetitive tasks such as adding inline tags, translators can focus their energies on crafting high-quality texts. Meanwhile, the term extraction and auto-aligner tools help ensure consistency of terminology, messaging and brand voice.
- Boosted productivity: XTM’s term-extraction tool allows terminologists to create glossaries up to 80% faster when compared with manual glossary creation, and creating well-structured translation memories via the alignment tool increases productivity by 90%.
- Reduced costs: The alignment tool also allows you to make better use of language assets, reducing the amount of time needed for human translation and cutting translation costs by as much as 90%.
Given the many benefits for productivity, cost and translation quality, it’s clear that linguistic AI is here to stay. But that doesn’t mean it’s staying put. My team is already working on more features and tools to make our clients’ lives easier. Projects in the pipeline include the automatic identification of non-translatables, automatic translation review and sub-segment translation memory matching. Each of these will help make localization even better and even faster.
As for where linguistic AI will take us in the years to come, I believe we’ll see further developments in the areas of content generation, predictive typing and the conversational AI used for dedicated virtual assistants. I, for one, can’t wait.”
Rafał Jaworski
Linguistic AI Expert at XTM international
Related articles
Check out these suggestions.
