Twitter Event Detection Reproducibilty Toolkit for ECIR2019

The toolkit consists of three modules -- Twistor (artifical Twitter stream simulator), the evaluation module, and the event detection modules.

Here are the steps to setup the project in Eclipse:

1. Create an eclipse project
2. Import the source code cloned from the repository
3. Add the libraries from the lib folder to your Java Build Path
4. Have a look at the package main for starting the different modules.

Twistor

Twistor is forked from the Twistor Repository, you can find a introduction and more details about it there. However, we wanted a repository with a self-contained reproducibilty toolkit and therefore added the source code to this repository.

Event Detection Modules

The event detection modules are based on the data stream management system Niagarino Repository. We included the Niagarino project as JAR for executing the two event detection modules Shifty and Log-Likelihood Ratio and the two baseline methods TopN and Random Events.

Evaluation Module

The evaluation module compares the resulting events from the event detection modules with the generated events description file from the Twistor module.