Agricultural risk models from information in text

Abstract

This project will develop models that will evaluate risk to agricultural products from textual sources. This information can be added to a 'local view' generated from sensors and other information sources located at the farm, to allow farm managers to make optimal decisions.This project will create graph based models from text. The models will not be for a single crop or agricultural product as it will allow a user to configure the model to represent the agricultural product that the user wishes to model. New information in the form of text (such as weather reports, news stories, government studies, etc) ordered in time is then presented to the model which then makes a predication of sentiment distribution of the effect of this information upon the crop or agricultural product. This information is then exported as time-series data which then can be added to a decision support system that can assist the farm manager in make optimal decisions. Ultimately, this information will be integrated with Scicrop mobile phone application that will allow access to this information from any location. (AU)