Twitter usage across industry: A spatiotemporal analysis

High resolution social media data presents an opportunity to better understand people's behavioural patterns and sentiment. Whilst significant work has been conducted in various targeted social contexts, very little is understood about differentiated behaviour in different industrial sectors. In this paper, we present results on how social media usage and general sentiment vary across the geographic and industry sector landscape. Unlike existing studies, we use a novel geocomputational approach to link location specific Twitter data with business sectors by leveraging the UK Standard Industrial Classification Code (SIC Code). Our baseline results for the Greater London area identifies Construction, Real Estate, Transport and Financial Services industries consistently have stronger Twitter footprints. We go on to apply natural language processing (NLP) techniques to understand the prevailing sentiment within each business sector and discuss how the evidence can contribute towards de-biasing Twitter data. We believe this research will prove a valuable surveillance tool for policy makers and service providers to monitor ongoing sentiment in different industry sectors, perceive the impact of new policies and can be used as a low cost alternative to survey methods in organisational studies.