19 Dec. 2017 — Updates
Well, this past year we said good-bye to our beloved Thing One and Thing Two (Brayden Arthur and Dustin Chang) who built the first iteration of linkedmodernisms.uvic.ca and established our quad-store database in the Virtuoso Conductor environment. Our first runs at machine learning did not succeed as spectacularly as we would have like (hey, we’re not Stanford!), but we learned a lot. Since then, a new team has come on board, led by David Mitchell, and including Marcel Bernard, Pooja Bhojwani, and Yudi Santos. As always, the fearless Belaid Moa, of Compute Canada, continues to play Yoda to our young Skywalkers.
The upshot is that we have figured out some great advances! We have begun to train the machine learning algorithm on dbpedia, and importing triples from there into our own quadstore so that we can query and visualize them more quickly. We’re also just on the cusp of launching at topic model browser borrowed from the Indiana Philosophy Ontology (inpho.cogs.indiana.edu) that will greatly enhance the navigability of the site. More visualisations are on their way, and we’re finally gearing up to do some real automated processing to generate richer and cooler data. Stay tuned!
13 Nov. 2016 — Updates
We’ve been working hard on updating the site, killing off bugs, and getting more data into the quadstore over the last several months. Some things that didn’t work before, such as searching by predicate, work well now. Also, we’ve just pushed a significant new batch of metadata to the quadstore, so that results are more and more robust.
Our next steps involve
- integrating the Topic Model browser developed by the InPhO group into our own site to allow sophisticated topic modeling — both naive and directed — of the whole content of the REM
- integrating our data with those of other trusted quasi-open databases such as InPhO and Linkedjazz to help fill in gaps in our data and vice versa
- simplifying and streamlining our machine learning process so that we can run it often, enhancing our data and results on a monthly basis at least
- building in a pedagogy annex where we will post unit plans, lesson plans, and sample assignments for the use of educators from high school through university
Stay tuned for more information and updates as we get them posted!
9 Jan. 2016 — IT’S ALIVE!!!
We’re thrilled to announce that Linked Modernisms is now live at linkedmods.uvic.ca. The metadata we are working with is a long way from complete as yet, but many, many of the sorts of things people interested in modernism might search are returning great results. Please
- Give it a test drive.
- Contact me if you would like to contribute content to help improve the site.
- Be patient — our machine reasoning and NLP training are ongoing and the data gets better every month.
- Let us know if anything seems broken or inexplicable.
Remember: it’s at linkedmods.uvic.ca !
9 Sept. 2015 — Update
We’ve done a huge amount of work in the last year, and plan to launch Linked Modernisms in the fall of 2015. Stay tuned. In the meantime, here’s the ontology created by Jana Millar Usiskin, Christine Walde, and Caroline Winter. It assembles aspects of several existing ontologies to create an interoperable mash-up that — we hope — will accommodate any and all aspects of cultural production. We call it Casaubon. Casaubon: XML version (in PDF) Casaubon (image) Millar Usiskin, Walde, and Winter are preparing a full-length publication describing the process and nature of Casaubon. Brayden Arthur and Dustin Chang, with the invaluable support of Belaid Moa (of Compute Canada), have worked with Casaubon to create an RDF quad-store in the Virtuoso environment. It features a near-natural-language interface for running SPARQL queries, and five visualization options. There’s a relationship finder, and some sophisticated searching capability allowing users to stack queries. It links directly to articles in the Routledge Encyclopedia of Modernism (going live this fall as well!). Finally, it uses both metadata harvested via questionnaire from domain experts and machine reasoning to discover entities in the REM, and to infer relationships among them. All three of Casaubon, the inference algorithm, and the content of the REM are dynamic, constantly evolving and updating one another to produce an ever-clearer picture of modernist cultural production the world over. (on dynamic ontologies, and the principles we have adopted for the Linked Modernism’s use of the REM, see Niepert, Allen, Murdoch, and Buckner at https://inpho.cogs.indiana.edu).
20 April 2014 — **I am extremely pleased to announce that Linked Modernisms (LiMo) has been awarded a SSHRC Insight Grant for five years (2014-2019)** Thanks are due for this success (already) to the project’s co-applicant, Susan Irene Brown (University of Guelph and University of Alberta), and its collaborators, Colin Allen (Indiana University, Ray Siemens (University of Victoria), Jon Saklofske (Acadia University), Christine Walde (University of Victoria), and Jentery Sayers (University of Victoria).
Building upon the Routledge Encyclopedia of Modernism (REM), Linked Modernisms (LiMo) will create an unprecedented resource for the study of global modernism. Its key objective is to enable the study of modernism as a truly global phenomenon. LiMo will make the world of modernist cultural production accessible to scholars by bringing Big Data methods to bear on a trans-disciplinary and trans-national body of scholarship. To this end, LiMo combines expertise in the fields of dance, film, literature, visual arts, architecture and design, performance, music, and intellectual currents with leading work on information ontologies, topic modeling, and faceted searching. With LiMo, a user will be able to explore global modernism in terms of characteristics or relationships instead of keywords. Instead of searching for entries that contain the terms Spanish, painter, and dancer, a scholar can search for all entries that are about Spanish painters who are dancers. The entries themselves need not contain the search terms, as long as the metadata does. Applied to the REM, this method will transform our understanding of global modernism and break new ground in computational modes of humanities inquiry: it will enhance what we know about global modernism by changing how we come to know it. LiMo will allow us to move beyond simply asserting modernism’s global dimensions to demonstrating it concretely, showing the vectors by which modernist memes, techniques, themes, approaches, materials, and ideas traveled and retraveled the world. Because the knowledge-base upon which LiMo is built takes in modernism in all its modes, and because our data-gathering efforts aim to be as comprehensive as possible, LiMo will become a crucial scholarly resource for the study global modernism.