Mainstream communication of big data using natural language generation (NLG)

Impact: Economic and/or Commercial

Description of impact

Research led by Professor Reiter and Dr Sripada developed natural language generation technology (NLG), which can generate English summaries of complex data. In 2009, this research led to (A) formation of spinout company, now called Arria NLG and (B) creation of an open-source NLG library, Simplenlg. In the 2014-2020 period, NLG technology was adapted by many more types of users (who did not use NLG before 2014). For example, Arria’s NLG technology was used by media companies (e.g. BBC) to generate news stories, professional services companies (e.g. Ernst & Young) to generate financial reports, and business intelligence (BI) users to get insights about data; and Simplenlg was used by the travel company Trivago to generate hotel descriptions and Bayerischer Rundfunk, a public-service radio and television broadcaster in Munich.

Outcomes to Date / Future Developments

Reiter and Sripada’s research led to the formation of a spinout company Data2Text, which was acquired by Arria NLG in 2013. Reiter and Sripada were seconded part-time to Data2Text and then to Arria NLG and became its Chief Scientists. In their Arria role, they were granted over 30 patents on NLG in the 2014-2020 period (https://www.arria.com/core\-tech/arria\-core\-patents/\). Most of these patents built on and extended the NLG technology and concepts, which they developed through research at Aberdeen University.

Since it was founded, Arria has raised over USD80,000,000 in investments to support NLG technology. Arria currently has offices in Aberdeen, London, Auckland (New Zealand), Sydney (Australia), and Morristown New Jersey (USA). It employs over 100 people, including 47 people in the UK (primarily in Aberdeen). The majority of Arria’s developers are located in the UK.

In the 2014-2020 period, Arria applied its NLG technology and made an impact on many new markets, including business intelligence, financial reporting, and automated journalism.
Impact statusImpact Completed (Open)
Impact date20142020
Category of impactEconomic and/or Commercial

Keywords

  • REF2014
  • Technological