Copy the page URI to the clipboard
Siddharthan, Advaith; Ponnamperuma, Kapila; Mellish, Chris; Zeng, Chen; Heptinstall, Danny; Robinson, Annie; Benn, Stuart and van der Wal, René
(2019).
DOI: https://doi.org/10.1145/3231588
URL: https://cacm.acm.org/magazines/2019/3/234928-blogg...
Abstract
We introduce and evaluate Blogging Birds, a novel artificial intelligence program that generates creative texts to communicate telemetric data derived from satellite tags fitted to red kites as part of a species reintroduction programme. We address the substantial challenge of effectively communicating telemetric sensor data in real time by (a) enriching this data with meteorological and cartographic data, (b) codifying ecological knowledge to allow creative interpretation of the behaviour of individual animals with respect to such enriched data, and (c) dynamically generating informative and engaging data-driven blogs aimed at the general public. Our evaluation found that summarising environmentally-enriched satellite-tag data as informative, engaging and fluent blogs was highly challenging even for trained ecologists, and that the computer-generated blogs were preferred by readers, notably where movements of focal animals were wide-ranging. We conclude that Natural Language Generation, and specifically data-to-text technology, has come of age sufficiently to achieve more than just the factual summarisation of data for professional use, thus opening up new avenues for addressing societal challenges related to communicating data effectively and engaging the public with scientific research.