Text and data mining of social media to map wildlife recreation activity

Abstract

Mining of social media has been shown to be a useful tool for social and biological research (e.g. tracking disease out breaks). This article outlines an accessible approach to the use of text and data mining (TDM) of social media to gather information on wildlife recreation activity. The spatio-temporal distribution of the shore based recreational European seabass (Dicentrarchus labrax) fishery in Wales is used as an example. Public online user generated content was mined using automated scraping. Data on fisher activity and fish sizes were extracted and then georeferenced by matching place names to a custom compiled gazetteer. Numbers of trips and spatio-temporal trends in the distribution of activity and catches were estimated. Prosecution was higher in summer than winter, and gear use and trip durations were consistent during the period 2002–13. Comparisons of TDM with existing surveys showed higher levels of activity and catch, and shorter mean trip durations were estimated using TDM. Monthly activity correlated closely with existing survey data. Spatial and temporal data agreed qualitatively with expert knowledge. This article showed that TDM can be used to describe a wildlife recreation activity, but use of TDM to derive unbiased population level estimates is challenging and more work is required to develop appropriate methods to correct for bias. These methods required no expertise in natural language processing or machine learning, a working knowledge of programming (e.g. in Python or R) is all that is needed to apply this approach. The opportunities to use TDM will increase with the continuing adoption of smartphones in emerging economies and developing nations and is of may be of particular utility where other data is unavailable.