This Company Uses NLP to Find the Right Keyword

3 min read

This is the ninth post in our series: Discover Korea’s Tech, where we will talk to a mix of Korean startup entrepreneurs who stood their own ground with their technology, in Korea’s economy notoriously dominated by gigantic companies. Stay tuned over the coming month as we talk to Korean entrepreneurs. You can follow our updates @technodechina for new stories in the series.

Entering a new market, especially one with a different language than yours, is not easy. All of your content needs to be expressed as naturally as possible. Finding the right words and expressions to accurately delivery your brand’s image and message is no easy task. Indeed, no matter how much time is put into crafting your content, fatal mistakes with awkward or incorrect translations are difficult to avoid.

That is exactly the challenge that Twinword hopes to solve by using natural language processing to make optimal language choices.

Since much of the Internet is based around search engine discovery, getting your keyword right is hugely important. Founded in 2012, Twinword combines data science and SEO/keyword research to deliver high quality keyword results for any business by extracting word associations and word relationships.

In the beginning of November, they released a new product called Twinword.Ideas.

“For those working in marketing industry in US are very keen to find relevant keywords and have had to pay a huge cost to advertisement companies for this role. Now, they can use Twinword.Ideas for that.” says Kono Kim, the founder of Twinword and a PhD candidate in natural language processing. On the other hand, Twinword also helps Korean and Chinese companies struggling to grab US consumers.

It is important to note that Twinword.Ideas is not merely a tool providing keywords.

“There already are a lot of startups providing similar service and Google Keyword Planner is doing this even for free”, says Kono.

Although these normal keyword research tools provides you with a long list of supposedly relevant keywords. But, this list is calculated purely from big data, it is very raw, leaving users to actually check one by one for relevance.

“Our technology provides other ways to play with that list. We do this through a technology called semantic sourcing which basically reprioritizes the order of words. We let the users catch the words that more specifically fits their demand and intent”, Kono said.
unnamed
There are two different features to realize this goal. The first one is User Intent; this filters keywords by the intent users have when searching between Know, Do, Buy, Local, and Web. For example, if the user chose ‘Buy’, only transaction-related keywords and buying-behavior related keywords will show up.

Another feature is Target Relevance. By freely entering the important element in keywords, the order of words changes to be more relevant. So, if I write ‘price’, then words that are related to price are put in higher ranks.

Considering that it has only been three weeks since Twinword.Ideas launched, it is impressive that Twinword.Ideas has been achieving more than 5 percent increase every week regarding user acquisition and traffic. And they have already established partnerships with US ad management solution platforms to integrate Twinword.Ideas keyword research tool.

 

“We are also close to partnering with SEO service platforms specialized in Chinese. So, in the future, as our existing clients want to expand their service to Chinese websites, we will be ready to provide that service as well”, says Kono. “Our goal is to assist companies in better understanding their consumers. We set up keyword strategy, utilize keywords on contents, validate traffic and results, and then at last, verify which keyword actually attracted their potential customers. It is a comprehensive cycle that we manage.”

Twinword is supported by K-ICT Born2Global Center, a major Korean government agency under the Ministry of Science, ICT and Future Planning (MSIP).

Image credit: Twinword