Quantitative analysis of crisis events as part of D.Phil project at Oxford.
These are considered works-in-progress.
The primary goal of this research is to develop methods by which to filter social media content that is generated during crisis events for messages which contain 'ground truth' information.
These notebooks represent research logbooks for the benefit of the author, and therefore are considered works-in-progress which are not intended as final pieces. For these, refer to the related publications or contact the author directly.
The data for most of this work was collected using a custom Twitter streaming app, written by the author, which collected live data during various crisis events. The live method allowed for detection of changing metrics (such as changes in follower networks) and collection of detailed user network data. The software is available in this repository.
Within the notebooks, data is pulled directly from the local Postgres databases
using Django syntax (via the Django shell_plus extension) as the data
collection software uses the Django framework to record its data.