By Associate Professor Rodney Clarke

The usual approaches to analysing the content of Tweets involve, for example, statistics, as well as techniques from the artificial intelligence community, specifically text mining and machine learning. Statistical approaches generally involve the analysis of word frequencies and the use of visualisation aides like tag clouds to indicate which key words are significant. Often these techniques are used in combination to examine the occurrence and collocation of key words associated with flooding.

As useful as these approaches are, they can only consider what was said; that is, they utilise only the lexical (wording) aspects of social media messages. There is a fundamental restriction in the use of these approaches. They are asemantic, that is, they cannot deal directly with the meanings expressed in these social media messages.

A novel but almost entirely unused possibility is to employ functional communication approaches that can be used to consider not only what was communicated, but also how a social media message is organised as an instance of communication. So-called functional communication approaches can be used to explore the grammatical resources that were used to formulate a given social media message.

This opens up an entirely new approach to the analysis of social media; one that can be used to describe the semantics of social media messages in emergency contexts. From a practical perspective, by analysing how social media messages in emergency situations are organized grammatically, we can exclude a large proportion of messages that are otherwise irrelevant. In effect, grammar provides many additional categories that can be used to filter, search and process the large collection of social media messages that can be collected during emergencies.

Associate Professor Rodney Clarke is Manager of the Collaboration Laboratory (Co-Lab) at SMART Infrastructure Facility.