[Humanist] 29.512 tools to analyze poetry

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Sat Nov 28 08:32:44 CET 2015

                 Humanist Discussion Group, Vol. 29, No. 512.
            Department of Digital Humanities, King's College London
                Submit to: humanist at lists.digitalhumanities.org

        Date: Fri, 27 Nov 2015 06:06:22 -0500
        From: margento <margento.official at gmail.com>
        Subject: 29.496 tools to analyze poetry Re: Humanist Digest, Vol 86, Issue 19

Hi Abraham and all,

Me and my team have worked in digital poetry analysis for a couple of years
now, have published a paper on multilable subject-based classification of
poetry and more recently have submitted (together with my team) a paper on
meter and rhyme to a DH journal and one on meter only to an artificial
intelligence conference (FLAIRS, where we presented the poetry subject
classification paper last year, here is the link:

http://webcache.googleusercontent.com/search?q=cache:7COEISzFQPUJ:www.aaai.org/ocs/index.php/FLAIRS/FLAIRS15/paper/download/10372/10322+&cd=2&hl=en&ct=clnk&gl=ca ).

Here is also the abstract of the forthcoming meter and rhyme paper:

Poetry Classification by Meter and Rhyme as Part of the Graph Poem Project

MARGENTO (Chris Tanasescu)
margento.official at gmail.com

Bryan Paget
bdjpaget at gmail.com

Diana Inkpen
diana.inkpen at uottawa.ca
University of Ottawa
August 25, 2015

Abstract: This paper presents a brief introduction to the the Graph Poem
project, which promises a novel approach for analyzing large corpora of
poetry. By treating poems and their shared features as nodes and edges of a
graph, we can study previously unknown relationships. To this end, we
repurposed an open source poetry scanning program (the Scandroid by Charles
O. Hartman) as a feature extractor for poetic meter and incorporated it into
our machine learning workflow. We also made our own rhyme detector using the
Carnegie Melon University Pronouncing Dictionary as our primary source of
pronunciation information. Initial work already shows a useful ability to
classify poems by use of poetic meter and rhyme. Future work will involve
assembling the graph depicting the interconnected nature of poetry (and
perhaps human thought in general) across history, geography, genre, etc.

Keywords: Machine learning, digital humanities, poetry

We also have a project websitehttp://artsites.uottawa.ca/margento/en/the-graph-poem/  at uOttawa.

Please let me know if you want to know more about the tools we've developed.
All best,


More information about the Humanist mailing list