NLG: A Separate Layer of the Technology Stack
August 12, 2019 | Greg Williams

Question: “Where does NLG sit in the technology stack?”

Answer: “As the push toward digital transformations gains momentum, maintaining Natural Language Generation (NLG) as a separate layer of the technology stack is ever more critical for ensuring the democratization of data literacy. Sure, Natural Language Query (NLQ) technologies can facilitate the distribution of BI beyond data scientists to less technically oriented employees, who for the first time are able to interrogate the data on their own, but without NLG there is no guarantee that these employees will understand the response.”

A data scientist can provide complex answers in un-narrated visuals that make sense to other data scientists. But for the rest of us, it’s hard to beat natural human language for conveying information.

What is a controller, sales manager, or head of recruitment to make of an un-narrated tree map or box-and-whisker plot? If a chart or data set can be explained in natural human language, it should be explained in natural human language. This is all the truer in a mobile Conversational AI environment, such as when the user is driving a car, because the machine’s spoken responses must stand on their own rather than as an accompaniment to a visual. In all cases, without NLG the possibility of misinterpretation or confusion remains high, as do the risks of abandonment, weak adoption rates, and stalled digital transformations.

Only an NLG platform with built-in advanced mathematical and language functions ensures that machines respond to NLQ with the power and clarity of natural human language, whether written or spoken. Working in tandem, the two complementary natural language generation technologies—NLG responding to NLQ—automate knowledge and the transfer of actionable insights from machines to humans.

As Arria demonstrated live at VOICE Summit 2019 on July 23rd, the two technologies are pairable today in support of dynamic, multi-turn conversations between humans and machines. During a well-attended breakout session, Arria’s COO, Jay DeWalt, and Chief Scientist, Ehud Reiter, unveiled the breakthrough use of Amazon’s Alexa as a presentation layer over a Tableau dashboard integrated with Arria's NLG technology. Our Head of Technical Integrations, Kapila Ponnamperuma, demonstrated the integration live, conversing and receiving insights in natural human language. The technology that the Arria team displayed at VOICE is highly transportable to BI platforms and to digital assistants such as Alexa, and further to lifelike human presentation layers, such as one of FaceMe’s digital humans, who can reinforce the insights with appropriate body language and facial expressions.

NLQ technologies can understand the question and map the question to intent.

And separately, Arria’s NLG provides the natural language response.

The democratization of data literacy stands to have a profoundly positive impact on decision making broadly within businesses. To achieve it, business leaders must concern themselves with more than how the human speaks to the machine and how the machine understands the human. That is only one side of the conversation. To humanize the other half, the machine’s response, requires a separate and distinct Natural Language Generation (NLG) layer of the technology stack.

Subscribe to the Arria Blog