Users Bring Real Value to Big Data, Machine Translation

You are here

Users Bring Real Value to Big Data, Machine Translation

Reprinted from Wired Cloudline

Online consumers around the world have an estimated spending power of $50 trillion, according to Common Sense Advisory. Yet English speakers represent about one-third of Internet users. To reach 98 percent of online consumers, a company must reach beyond the English language and translate content into 48 languages.

Big Data compounds the challenge by flooding businesses with content from around the globe. The volume of global online business data doubles every 1.2 years, according to an academic estimate from the W.P. Carey School of Business. IDC projects that by 2020, there will be 450 billion business transactions taking place per day. Roughly 2.7 zetabytes of data already exist online today, according to an IBM estimate.

It’s no wonder American companies wishing to globalize face significant hurdles when it comes to translation. Not only do they have to translate content into dozens of languages, but they have to manage and prioritize their flood of global information. Only when companies combine cloud-based Big Data and translation applications with intelligent human users can they begin to harness this diversity of data to their advantage.

 

Big Data’s Unique Challenges

Big Data applications sort and analyze petabytes of online information. When correctly interpreted, Big Data statistics and trends offer enterprises new insights into their operations and marketplaces. But without people to give it meaning, Big Data is nothing but an excess of numbers.

Companies are still falling short when it comes turning Big Data into real business value. The onus is on humans to improve their interpretation skills. A recent Avenade survey found that more than 60 percent of companies need employees to develop new skills in order to gain real insights from Big Data. As Michael Schrage wrote in a Harvard Business Review blog post, “too many organizations don’t quite grasp that being ‘big data-driven’ requires more qualified human judgment than cloud-enabled machine learning.” Applications, no matter how intuitive or helpful, still don’t supplant human savvy.

Translation and Big Data: A Pandora’s Box?

The fact that companies must deal with dozens of languages and cultural nuances compounds the Big Data challenge. Professional translators, traditionally used for content translation, are in short supply. Moreover, human experts generally translate 2-3,000 words per day, according to Common Sense Advisory, and that number has been steady for decades. In a Big Data-driven world, enterprises potentially need to translate millions of words per day.

Machine translation is an old standby for generating faster translations. If you’ve ever seen an online translator spit out a phrase that doesn’t make sense, however, you know that machine translation can be hit-or-miss. This is because translation memory (TM) has traditionally been held in information siloes. TM refers to the database of sentence fragments and phrases that machine translation software uses to find the right translations. For a long time, companies stored their TM in private databases. As a result, machine translation only had a limited number of phrases to draw upon, resulting in fragmented translations. Enterprises also used different kinds of software to store and generate translation memory, so there was no uniformity across the board. For example, many companies wasted time by using a Computer Aided Translation (CAT) model, which only really worked form domain-specific content like weather reports and legal documents.

The Ultimate Database: The Cloud

Today, companies are just beginning to migrate away from their old, inefficient translation technologies to cloud-based platforms. The cloud offers access to millions of human translators who can build up TM quickly, resulting in better machine translation. Cloud-based Big Data applications generate similar advantages, compiling information from all over the Web to help humans generate new insights. On top of that, companies experience the cost- and productivity benefits of virtualized infrastructure.

Like Big Data, however, the cloud by itself doesn’t solve companies’ most pressing problems. Namely, how do we translate huge amounts of contents quickly, without sacrificing quality? How do we generate real business value with Big Data? How do Big Data and global content work together to make our business smarter?

Human, Meet Machine

The answer lies in humans as intelligent end users. People who can translate Big Data statistics and trends into actionable information remove the disparity between numbers and business value. Likewise, people who take raw machine translations and edit for idioms, colloquialisms and other cultural nuances will overcome the hurdle that separates bulk translation from good translation.

The benefits of adopting a new, human-meets-data way of thinking will expand as Big Data and globalization takes a firmer hold on business culture. As trends are spotted in certain regions, businesses can react in real-time with targeted translation efforts, adjusting over time based on further trends. Likewise, as people continue to feed translation memory, machine translation will generate better results, enabling businesses to expand more quickly and easily in their global markets. The enterprise world is headed towards a Big Data and machine translation-driven future — and only people can bridge the gap.

banner block

Contact Me

Want to learn more? Contact us. It only takes 5 simple questions to determine if Lingotek is the right solution for your company.

banner block

Our Featured Customers

What makes Lingotek unique is our track record of building, powering and managing the most successful and vibrant brands in the world.

Latest Tweets