Live language analysis of Earth Hour tweets: how are people talking about Earth Hour?

by Diana Maynard and Kalina Bontcheva,
the University of Sheffield

Collective awareness about climate change is an ongoing problem because there is such a wealth of information available, which can be confusing, contradictory and difficult to interpret. In order to help citizens understand environmental concerns, and to help organisations better inform and target interested people with campaigns, we have developed an open source toolkit to analyse social media data on the topic of climate change. The toolkit comprises methods for extracting, aggregating, and visualising actionable knowledge, based on automatic analysis of large volumes of text.

The key terms, poster types and sentiments expressed in online discussions are extracted, along with key indicators of climate change, and are stored in a semantic search tool, which enables complex real-time searches over large volume social media stream.

During the Earth Hour 2016, we collected and analysed in real time all tweets related to the events, extracted all this information automatically, and posted tweet updates showing aggregated statistics on the types of language being used in English Earth Hour tweets, opinions expressed towards the event, and the types of users posting on the topics.

You can see some of the results in our demo here.