A little over a year ago, I announced that Zoltan Majdik, Dave Clark, and I received a NEH Digital Humanities advancement Grant. The overall aims of the Transparency to Visibility (T2V) project were to develop a toolkit that can aid humanities researchers by providing an easy-to-customize automated framework for converting unstructured text into nodes and edges, capturing the relationship among nodes using a combination of Named Entity Recognition, machine learning, and regular expressions. As a test data set, T2V used conflict of interest statements in medical publishing; these statements are only minimally structured, but contain relationships among writers and agencies that, while obvious to human readers, can be a challenge to capture in a database. As the active development phase of this project comes to an end, I’m happy to announce the following updates:
A new Conflicts of Interest Network Explorer Dashboard is now available. The app allows you to explore conflicts of interest networks in biomedical publishing. Documentation is available at:
http://conflictmetrics.com/documentation.
https://gitlab.com/grahamss/transparency2visibility
A detailed whitepaper documenting design, development, and user testing is now available. “Extracting Relational Network Data from Unstructured Texts: Lessons from the T2V Project” can be accessed at:
https://gitlab.com/grahamss/transparency2visibility/-/blob/master/graham-et-al-t2v-whitepaper.pdf