Skip to navigation Skip to content

Enhancing the provision of environmental information through machine learning

River users may benefit from NLG information. Photo: Spencer Gurley via Pexels
“More collaboration between public authorities, users and researchers is necessary to explore further applications of digital technologies, with social and natural scientists, to improve the provision and use of environmental information”

Public authorities are enabling citizens to be better stewards of the environment through greater access to environmental information, but there is a need to improve online information to meet people’s needs.

A new open-access paper involving the James Hutton Institute and led by University of Aberdeen researchers has found that natural-language generation, a machine learning process able to produce sentences and longer texts based on numbers or words that is used to produce, for example, written weather forecasts, can enhance the provision of environmental information for public use.

A study on the use of river level information, collected by the Scottish Environment Protection Agency’s national network of gauging stations along 232 Scottish rivers, showed that the same information was being used in entirely different ways by groups interested in fishing, flood risk, kayaking and canoeing.

Researchers worked with SEPA and users of their online river level information to experiment with the automated conversion of river level data into textual messages. Users found the generation of straightforward messages from changes in river level to be particularly useful to their specific purposes.

Dr Kit Macleod, based within the Institute’s Information and Computational Sciences group and co-author of the study, said: “Working with public authorities who provide environmental information and people who use it, can help us improve the provision of information. Natural-language generation could be widely used for more effective online environmental information provision.

“We found that profiling user groups and tailoring the information provided are important steps towards more effective communication, and there is significant value to a wide range of individuals and groups in the provision of straightforward messages, for instance using natural-language generation.

“More collaboration between public authorities, users and researchers is necessary to explore further applications of digital technologies, with social and natural scientists, to improve the provision and use of environmental information.”

Dr Rupert Hough, leader of the Institute's Information and Computational Sciences group, commented: “We live in a data and information-rich world. Without appropriate interpretation, this wealth of information can be obsolete.

“Kit and colleagues have developed an approach to providing tailored interpretation of complex data for multiple user-groups in an efficient way. There is significant potential to use this system in a wide variety of contexts.”

The open-access paper Towards more effective online environmental information provision through tailored Natural Language Generation: Profiles of Scottish river user groups and an evaluative online experiment is in the latest issue of Science of the Total Environment.

Press and media enquiries: 

Bernardo Rodriguez-Salcedo, Media Manager, Tel: +44 (0)1224 395089 (direct line), +44 (0)344 928 5428 (switchboard) or +44 (0)7791 193918 (mobile).


Printed from /news/enhancing-provision-environmental-information-through-machine-learning on 09/06/23 10:07:05 AM

The James Hutton Research Institute is the result of the merger in April 2011 of MLURI and SCRI. This merger formed a new powerhouse for research into food, land use, and climate change.