“I’m hungry. Let’s get tacos.”
What does that mean? One option is I am looking for a specific Reservoir Dogs clip. More than likely, though, I am in search of a place for lunch. What happens if you enter the phrase into a search engine? (Hopefully, you’re not very hungry.)
When you see results like these, your search experience immediately goes down the tubes. Why shouldn’t it – you expect a certain level of relevance from your search results. You want a search engine that provides results based on what you mean, not literally what you type or say. Unfortunately, the latter is what we have come to expect from most of our search engines.
This can be particularly devastating in customer service environments, where you might be trying to explain a problem you’re experiencing, but don’t have the technical nomenclature, so you do the best you can and hope for the best.
Inbenta doesn’t think you should have to hope for the best, but should expect it, and is working towards a new breed of search, based on meaning rather than keyword, using meaning text theory.
“The point of meaning text theory is what is important is not the meaning of every word taken individually, but the context of the combination of words,” explains Jordi Torras, founder and CEO of Inbenta.
When customers go to a website to seek assistance, they are often faced with the situation above, having limited vocabulary to explain a specific problem to which they are seeking resolution. Instead of being able to successfully use available self-service options, those customers become frustrated and eventually find themselves calling or emailing customer service.
“Our main target is to use natural language processing to avoid unnecessary calls and emails, while still offering great services,” says Torras.
What Inbenta does is take a set of FAQs from its customers and try to effectively predict any permutation of the question that might be used to make the same inquiry – and even other seemingly irrelevant questions – and pair them up with the right responses. It’s actually the opposite of an FAQ – inbenta tries to anticipate the most unusual questions and provide a relevant response.
“The problem for businesses is there are a tremendously high number of potential questions that might be asked, versus a very small number of relevant answers – that is why we search based on meaning and not keyword,” Torras adds.
In addition, there are likely to be multiple responses that leverage the same keyword, making it difficult to determine the most appropriate response without contextual analysis. In fact, Inbenta analyzes the full spectrum of search requests and groups them into clusters by meaning, helping define new content for its customers. This is particularly beneficial with long tail searches, which are commonplace today and much more difficult to understand based on keyword alone.
So the next time you enter a search phrase on a website, be aware that, unless the company is using inbenta, its search engine isn’t as smart as you are and you may need to search down to its level.
For you Reservoir Dog fans, here’s the clip with Mr. White and Mr. Orange.
Group Editorial Director
No matter what industry you work in, you've likely been hearing about the importance, and prevalence, of machine learning and analytics. But what do t…
With several announcements made during Apple's annual WorldWide Developers Conference (WWDC), here are the top 10 that mobile-first businesses should …
In tech circles and beyond, AI is the mot du jour lately, often thrown around in speculative conversations as the magical key that will unlock previou…
It used to be that news about DDoS attacks was largely limited to tech websites and other specialized information sources, where the focus was on atta…
According to Investopedia, algorithmic trading already comprises 70 percent of daily trading. As trading becomes more automated, the need for human an…