G-Flow is an automatic extractive summarization system that seeks to balance coherence and salience. G-Flow introduces a joint model for selection and ordering that balances coherence and salience. G-Flow's core representation is a graph that approximates the discourse relations across sentences based on indicators including discourse cues, deverbal nouns, co-reference, and more. This graph enables G-Flow to estimate the coherence of a candidate summary. G-Flow was developed at the University of Washington's Turing Center.


The following people have contributed to G-Flow:


You can download the summaries we used in our evaluation here.


For information on the G-Flow code, email Janara Christensen (janara AT cs.washington.edu). G-Flow is released under an academic license. For instructions on how to run G-Flow or use it in your own code, please see the README file (also included in the download).


Try out our system using our demo.


  • Hierarchical Summarization: Scaling Up Multi-Document Summarization
    Janara Christensen, Stephen Soderland, Gagan Bansal, and Mausam
    Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (ACL 2014)
    pdf | bibtex
  • Towards Coherent Multi-Document Summarization
    Janara Christensen, Mausam, Stephen Soderland, and Oren Etzioni
    Proceedings of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL 2013)
    pdf | bibtex