In the domain of social networks analysis, like Twitter, exploiting and analyzing data generated by users can give valuable information and insight about public opinion regarding an issue, detecting an event, or analyzing business information. As mentioned, one way of exploiting twitter data is to detect events based on the information being reported by users. For this purpose, burst detection approach is used to detect events.
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.
To use event detection code provided in EvenentDetection.py, you need to install Numpy. In addtion, having cloned the repository, unzip the fileclusters.sortedby.time.csv.zip. It is includede of 7 days (about 4 millions) twitts.
To run the code, the script EventDetection.py should be given an argument to determine number of windows. The script can be run like below:
python EvenentDetection.py NUM-OF-WINDOWS
The script uses the file fileclusters.sortedby.time.csv as the input. Thus, before running the script put the code and file in the same directory. The results will be stored in a file called NUM-OF-WINDOWSwind.csv.
To see the report on the results, the script eval.py is used:
python eval.py NUM-OF-WINDOWSwind.csv
Please read Event_Detction.pdf for details on the implemented method and the report.