For all the fantasy sports enthusiasts out there, have you ever thought about the emails you get from your leagues at the end of each week with write-ups about your team’s performance? If you think someone sits there and manually sorts through your data and writes each of them, think again. That would be an impossible feat.
It’s all possible using natural language generation (NLG) to create descriptive narrative around the data and tell the story in an easily digestible format. NLG uses artificial intelligence to turn structured data into common language to communicate insights and information from the data – exponentially faster than through human analysis. Using NLG also take human error out of the equation, adding a level of certainty to the analysis.
“The tech has not been available to truly get all the value out of the data that is available to us,” said Arria CEO Sharon Daniels. “We are generating so much data, but until you truly have precise written summaries – either by human analysts or technology – you just don’t get the information in a timely manner. In many cases, by the time you get the information, if you can get through it all, it’s old.”
Arria has built integrations into the top BI dashboards to simplify deployment, but the data can rally come from anywhere, including custom platforms or web apps. The NLG technology sits on top of wherever the data resides to tell the same story that is delivered graphically in an easy-to-consume format.
What’s even more exciting and will extend Arria’s NLG capabilities to a much broader audience is its new integration with Excel. Now, any data that is hosted in Excel databases can also be analyzed by Arria to churn out a narrative instantly. That means any business storing data in Excel databases can leverage NLG to tell their story instantly, bringing what has perhaps been an enterprise solution to all businesses.
Businesses don’t have a problem collecting data. They have a problem creating the story from their data. Having once managed surveys for a financial publication, I can attest to how long it takes to analyze data and put it into a digestible format. With Arria’s NLG, it would have been a simple as mapping the data to be used for the output and generate a report with the click of a button. What took days or weeks would have been accomplished in seconds.
Daniels recalled showing the technology internally to a finance team member, whose initial reaction was, “Well, there goes my job.” But, she quickly realized the report writing is the part of the job financial analysts, accountants, and others dislike most, and the NLG technology would allow them to spend more time crunching numbers and not having to worry about getting reports written in time. “This is going to make me look like a star,” she concluded.
When it comes to data-driven reporting – especially financial reporting – one of the biggest challenges is changes in the data. Inevitably, some data point changes last-minute, which results in an entire report having to be re-written and every number re-verified to account for the single new data point. With Arria, the report can be instantly rerun to account for the change with the knowledge all other impacted data have also been accurately reflected.
Arria’s API library extends the technology into countless use cases and applications across industries, and its pre-built library of many common analytics functions – trends, correlations, standard deviations, etc. – turning data into narrative can happen right out of the box.
For a very timely example of how Arria’s NLG is helping deliver current information, take a look at the COVID-19 map project Arria worked on with Microsoft using data from Johns Hopkins University. By clicking on any country, the site will generate a narrative about the COVID-19 status in that country or region. Or, you can drill down into individual states in the U.S. to see narratives about not only those states, but a county-level breakdown as well. You can also click to see the supporting tables used to generate the narrative, if you’re so inclined.
Arria also worked on an interactive dashboard with TIBCO.
But, what has Daniels really excited is the prospect of really changing how we interact with data and making it much more human-like with conversational AI. With its Arria Answers product, the company is bringing all the different natural language technologies together. Arria Answers is a conversational AI platform that integrates with BI dashboards or Amazon Alexa so you can ask your data a question.
The big breakthrough that Daniels thinks is going to draw attention is the true multi-turn drilldown capability, which she says nobody has truly been able to produce previously.
“You can ask Alexa a question, but it’s really just a very fancy search engine,” she says. “We have fun with it, but we can also become frustrated with it when Alexa can’t find the answer or gives a silly response.”
It’s not that the information isn’t always available. Rather, it’s that Alexa doesn’t always understand what to do with it or how to find it. In fact, more often than not, the information is out there – Alexa just hasn’t been trained to access or interpret it easily without multiple iterations of questions. Arria Answers is changing that.
“What Arria is doing is turning that information into language that can be spoken back through Alexa. Because we have multi-turn drilldown capabilities in our advanced NLG reporting, we can use these human-like speech interaction capabilities to improve the experience.
Multi-turn drilldown helps create contextual continuity and turn digital interactions into more human-like experiences. For instant, Arria Answers remembers what you have said, allowing you to refer to labels in a previous question and remembering where you are with your data, eliminating the need to repeat questions. This ability means users need only specify the relevant or new part of a question or request, and Arria will complete the unspecified elements automatically.
For instance, if you’re asking Alexa to tell you what your investment portfolio looks like today, it can tell you. But then, you can simply ask, “What was it like yesterday?” or “What caused the change?” and Arria will maintain the context of the conversation, just as if you were speaking to your financial advisor.
Arria’s NLG technology is all about explainable AI and data storytelling. Businesses are collecting so much data today, which could be valuable – if they were able to not only parse it and analyze it, but turn it into an easy-to-understand format. NLG not only takes that burden away, but makes that digestible format available in seconds, so information is timely and relevant and in users’ hands when they need it. With Arria Answers, those same stories can now be told verbally, as well.
Edited by Erik Linask