Natural language processing of biodiversity policy documents

Aims and Objectives

We aim to use the YESI Fellows scheme to assess the extent to which countries of the world are (or are not) considering the inter-related challenges of biodiversity conservation and climate change in a holistic way. The topics of climate change and biodiversity overlap in a range of ways, and so countries can benefit greatly by ensuring that their respective strategies speak to each other coherently. Currently, however, it is unclear whether governments are even acknowledging the fact that these relationships exist, let alone seeking to purposefully align their actions and targets.

We will develop and apply a bespoke index that conveys the extent to which biodiversity strategies are integrating matters of climate change, and vice versa. Findings can then be used to compare across documents to gauge the degree of alignment between both, and to assess whether this has improved over time. Traditionally, content analysis such as this would involve manually reading and coding the relevant text in each document, and we hope to bring novelty and efficacy by applying natural language processing (a form of machine learning) to the task.

As well as highlighting ways that countries could improve their environmental strategies, the outputs from this work will have future applications. For example, the annotated texts could be used to assess alignment between the environmental strategies of local administrative units in a single country. Moreover, given that countries produce new strategies roughly every five years, we hope that this project will mark the beginning of a longer-term effort to monitor and inform environmental planning processes into the future.

Outputs

A Hybrid Question Answering Model with Ontological Integration for Environmental Information. Tianda Sun, Jamie Carr and Dimitar Kazakov. Accepted peer-reviewed paper.