This python code:
- builds a word graph starting from a set of tweets
- allows updating the word graph with new tweets at any time
- answers summarization requests at any time
This code was developed to experiment with summarization techniques for the paper "Efficient Online Summarization of Microblogging Streams". This is academic code. It is duct tape code. It is not production code. Please use it only to understand concepts such as word graphs or lazy-updated decaying windows.
Tweets are not included since this is not allowed by the Twitter TOS. The dataset can be reconstructed by querying the Twitter API using the tweet ids provided.
If you have any questions about this code, please contact me [1] or check out the paper at EACL 2014 [2].