Margaret Warren

Visiting Research Associate

Margaret Warren is a researcher at the intersection of semantics, art, metadata, and information science. For most of her working life, she has been deeply interested in how people describe what they experience visually and how those descriptions can be formalized into structured machine-readable data.

Margaret was born to artist parents, so being an artist was not only encouraged, but pretty much assumed to be a part of her DNA. However, early on she was also interested in electronics and technology, so she officially launched her working life with a tour of duty in the U.S. Coast Guard as a Cryptographic Technician. The Coast Guard led to studies at Florida State University in Computer Science and jobs in multimedia and technology. Before forming her own company to provide networking, hardware, and software services across the Gulf Coast (which she successfully operated for over 15 years) she held jobs in software programming, network system administration, and at one point, a contract with the State of Florida, where she educated the House of Representatives staff in their computing environments.

They say life is what happens when you make other plans. Just as Margaret was embarking on focused academic research into art and semantics, her fiancé unexpectedly passed away. In managing his estate, which included running his vintage Porsche restoration company for over a decade, she also learned many new fun things such as how to build early air-cooled Porsche motors and program electronic control units for fuel-injection modified Porsche 356 engines. She also completed a BS in Information Technology from the University of West Florida during this era and formed a third company: Metadata Authoring Systems (MAS), for the purpose of pursuing her research interests in earnest. Through MAS, and with some involvement from researchers and programmers associated with Florida IHMC and a German company called Netestate, Margaret has created a data-centric, human-machine interface called ImageSnippets, which is used for image annotation research.

This fully operational web application empowers archivist and curators, who can use it to annotate and publish images on the web with industry standard structured linked-data. Perhaps more interestingly, however, it functions as a system for ontology engineering based on visual prompts for subject matter experts. In this mode, it is a pioneering human-in-the-loop interface using AI-augmented image labeling for knowledge graph construction and can be highly useful for training AI models in scene graph prediction. The included search module, which is based on a custom ontology of relations, functions as a novel way of tracing semantic explainability over knowledge graph inference graphs using visual cues.

The system has been the subject of invited talks at Google, IHMC, Florida Institute of Technology, the Internet Archive, and the CEPIC/IPTC Photo-Metadata conference, among others. Margaret holds a U.S. patent on a key aspect of computable semantic annotation and has co-authored publications for the Second International Conference on Concept Mapping in Costa Rica, 2006; HASTAC (Humanities, Arts, Science, and Technology Advanced Collaboratory), Duke University, 2007; the AAAI Spring Symposium, Stanford University, 2010; and the 1st Workshop on Subjectivity, Ambiguity and Disagreement in Crowdsourcing, HCOMP, Zurich, 2018.

In July, 2019, she was part of a small team that built an experimental, functional mesh network for the Decentralized Web Camp, Pescasdero, California sponsored by the Internet Archive, where she also gave talks on the creation of image graphs and the importance of using structured metadata for media integrity on the web. She is an active member of the IPTC (International Press and Telecommunications Council) photo-metadata working group, which is responsible for developing standards for image metadata in digital media resources. Additionally, she is a member of the DBpedia Association, a large-scale public knowledge graph that sits at the heart of linked data on the web. In 2020, she chaired a track for the DBpedia sponsored conference, “Knowledge Graphs in Action”, where she gave an invited talk, “Creating and Using Image Graphs: Recent Research in AI Augmented Construction and Using Images for Exploring Inference Graphs.”

As an artist, Margaret has won awards in juried art shows, and her work has been used in book design and purchased for collections in Greece, Sweden, and around the United States. Her painted Porsche 356 ‘art car’ won an award at the 2009 Houston Art Car Parade and has been shown at the Glenmoor Gathering of Automobiles, Porsche 356 Registry events, and the Amelia Island Concours d’Elegance, as well as several museums and galleries throughout Florida and the eastern United States. She is a founding member of the SAP artist collective, has been a studio artist at First City Arts in Pensacola, Florida, and a visiting artist at Developing Environments in San Francisco, California, Summer 2019.

Margaret’s research interests include knowledge representation, ontology engineering and the decentralized web. She is passionate about media integrity, valuing human content and data creators – whether they be artists, subject matter experts, or gig workers in the data economy.

Besides art and semantics, Margaret also enjoys yoga, meditation, working in pit crews at Sports Car Club of America (SCCA) racing events, long road trips, hiking, and Formula 1 racing. Her favorite thing to do is create (usually) poorly understood, conceptually complicated art projects.