From Tweets to farm management: Mining agricultural information and discovering agricultural communities in social networks

Abstract. New opportunities for farm management decision making have been rapidly growing with the proliferation of data and information describing agricultural systems. This is a stark contrast to the recent past when farmers would have access to a comparatively limited number of sources, with high latency, when obtaining data for decision-making. This could lead to sub-optimal farming decisions and diminished system performance. Having the ability to obtain real-time information about the occurrence of culture tasks within their region may help farmers make local decisions for their farms. Based on our preliminary studies using text mining on news media, we saw potential concerns with high latency, particularly when several days of delay in information delivery to decision-makers occurs. In this study, we established a new pathway for acquiring agricultural information in real-time utilizing Twitter, which also provides geolocation data with finer spatial resolution. This allowed us to extract information for identifying the actual timing of local crop planting schedules. We have also identified influential agricultural information providers within social networks, based on social network connections of the communities observed within Twitter. The results have the potential to not only enhance farmer decision-making, but also facilitate discovery of information flows within agricultural communities.  This will provide new strategies for the development and deployment of targeted community learning modules for enhanced implementation of best management practices.