In the APSA Public Scholarship Program, graduate students in political science produce summaries of new research in the American Political Science Review. This piece, written by Lizzie Martin, covers the new article by Baekkwan Park, East Carolina University, Kevin Greene and Michael Colaresi, University of Pittsburgh,“Human Rights are (Increasingly) Plural: Learning the Changing Taxonomy of Human Rights from Large-scale Text Reveals Information Effects”
Tracking human rights abuses is key to addressing them. Unsurprisingly, advances in technology have made discovering rights violations easier. Access to smart phones has increased coverage of human rights on social media, and advances in satellite imaging have created more visual data. However, these advances can make it harder to identify changes over time consistently: they can leave human rights advocates wondering whether abuses of human rights are increasing, or whether the information available about them has increased. This potential confusion makes addressing urgent problems even more complicated. Political scientists Park, Greene, and Colaresi meet this challenge with a new framework for understanding human rights and new methods to track them.
Traditional approaches to tracking human rights abuses over time often fail to account for changes in technology and thus information. Some track the word counts of reports on human rights, assuming that longer reports correspond with more concerns. Others have scored reports, relying on subjective assessments of rights violations. Neither method considers the way that increases in information might change the way that reports are written or how new technologies might change the way we understand rights themselves. We generally think of human rights as a group of concepts, such as various political rights, protections from discrimination, and religious freedoms.
For example, concerns about civil liberties in a country may have been captured in broad discussions of issues like freedom of speech and movement in early reports. As technology evolved, more information about the ways these rights were violated might have become available in the form of social media posts about limitations on speech or images of border checkpoints. Additionally, new rights themselves might have emerged with advances in technology, such as online speech. This influx of information and broadening of potential rights to consider could make reports longer, which traditional approaches would interpret as an indication of more severe rights violations, even if the abuses that the reports described remained materially the same.
As a solution, Park, Greene, and Colaresi propose a new conceptual representation of human rights. We generally think of human rights as a group of concepts, such as various political rights, protections from discrimination, and religious freedoms. However, the authors suggest that we instead consider them as a structured taxonomy, similar to a family tree, with broad concepts that branch into more specific, nested rights. Civil rights might branch into freedom of movement, assembly, religion, speech, etc., each of which might also branch into more specific categories of their own. Changes in the availability of information can make tracking rights abuses difficult
This new way to think about human rights provides a way to distill vast amounts of evolving information systematically. These taxonomies, or trees, are easier to compare over time than rights reports. In a country with longstanding abuses of labor rights, for example, that broader category would appear even in early versions of its rights taxonomy. Those tracking rights violations could see when new information simply allowed future reports to fill in more specific branches underneath that concept, such as through providing details on working conditions, rather than indicating that abuses are materially worsening over time. Based on this new understanding of human rights, Park, Greene, and Colaresi apply tools from machine learning and natural language processing to reports from the U.S. Department of State, Amnesty International, and Human Rights Watch. They find that as information has increased, these reporting agencies’ maps of rights have gotten deeper and more complicated. In addition to illuminating how new information can change our understanding of human rights over time, this approach allows the authors to identify new rights concerns that have emerged.
Changes in the availability of information can make tracking rights abuses difficult. Park, Greene, and Colaresi’s work answers that challenge. Their approach makes it possible for advocates working to fight or prevent human rights violations around the world to distill a potentially overwhelming influx of new information into a meaningful understanding of how rights abuses have changed – or not – over time.