For many important nature conservation programmes, western societies are increasingly reliant on the activities of volunteers, who, collectively, have come to represent an unpaid work force of considerable size and importance. Although a variety of effective ways exist to generate adequate recruitment, volunteer retention is harder to achieve, particularly when schemes grow bigger or tasks get more difficult. We describe two case studies that we are using to investigate the hypothesis that richness of information provision, of the kind that can be provided by Natural Language Generation (NLG), can play a role in fostering volunteer interest and motivation. Both these case studies involve collaboration with large existing conservation projects, which provide the possibility for evaluation on a realistic scale.
Can narratives that bring to life the behaviour of specific members of a species have a positive effect on the involvement of citizens in conservation activities involving that species?
We are investigating the above question by building an NLG system that is able to produce a commentary on the travels of red kites that are tagged with satellite tracking devices. The red kite is a species reintroduced to the UK, whose success varies in currently unexplained ways across regions. For this project, we are working with one of the largest nature conservation charities in Europe. Our aim in the red kite project is to bring these tagged birds of prey to life by constructing narratives around their daily activities. Currently, the satellite tag data is being used by the charity to manually create blogs such as:
"...Ruby (Carrbridge) had an interesting flight down to Loch Duntelchaig via Dochfour on the 6th March before flying back to the Drumsmittal area, spending the 10th March in the Loch Ussie area (possibly also attracted by the feeding potential there!) and then back to Drumsmittal for the 13th..."
Our interest, in addition to automating the generation of such blogs is in making these narratives more interesting, by using the data to illustrate key aspects of red kite behaviour.
Can volunteers learn identification skills better if they receive tailored feedback than if they receive fixed feedback?
Citizen science projects aim to collect vast quantities of data from untrained volunteers. In the ecological domain, the most successful of these ventures is the RSPB's big garden birdwatch (http://www.rspb.org.uk/birdwatch), which attracted over 600,000 volunteer recorders this year. Given that the common garden birds in the UK are easily identifiable, such volunteer collected data can give an accurate picture of trends for some species. However, for less common species, or species that are harder to spot or identify, such citizen science efforts are less reliable and also attract and retain fewer volunteers.
In this project we are attempting to use NLG to increase volunteer motivation and volunteer accuracy in identifying different species of bumblebees. There are 25 different species of bumblebee in the UK; many of these are threatened by habitat loss. This is a matter of concern as bumblebees are important pollinators of wild and garden flowers as well as crops and fruit trees. Bumblebee identification requires some training and, particularly when identifying from photographs, can be quite challenging. Thus, both volunteer motivation and accuracy are key issues for a citizen science project in this domain. The bumblebee charity that we work with runs a scheme where volunteers submit photos of bumblebees through a web interface, which are then identified by an expert. This approach doesn't scale and the question is whether we can train volunteers to accurately identify the bees, by giving tailored feedback in this process. We have two aims:
- To provide useful feedback when users make an incorrect identication, to improve identification skills.
- To use "crowd-sourcing" techniques to reduce the effort required from the domain expert.
Next page: People