This project brought together an interdisciplinary team of researchers in history and philosophy of science, computational humanities, biology, and the arts. The central topic of the project was how ambiguity can be productive in scientific research, focusing specifically on ambiguity in taxonomic classifications. The project investigated how the history of taxonomy can shed new light on the value of ambiguity for science in the domain of data-intensive science, complementing existing research on figurative language in science. The project focused in particular on detecting trade-offs in the value of ambiguity for scientific language as a function of changing social contexts. The broader theme of productive ambiguity in the setting of computational data processing has contemporary relevance far beyond current challenges in biology: the idea that we should eliminate ambiguity wherever possible remains entrenched in many efforts to develop controlled vocabularies or ontologies for the Semantic Web today, including in the digital humanities and library sciences.

Despite disruptions to the team’s ability to collaborate during the Covid-19 pandemic, the project achieved major outcomes in each of main areas listed above: it synthesized prior interdisciplinary knowledge about productive ambiguity to demonstrate and evaluate its significance for science, created innovative computational methods for measuring ambiguity in scientific texts, and created new courses and teaching materials engaging ambiguity as a theme at the intersection of art and science. The project has supported eight peer-reviewed publications in history and philosophy of science, artificial intelligence, and biodiversity data science. Several more manuscripts are currently under peer review at journals in the fields of knowledge representation and scientometrics. Results were also disseminated through presentations at leading international conferences in the fields of philosophy of science and biodiversity data science. PI Sterner is continuing to develop methods and empirical results created during the project through new and ongoing collaborations.

Broader impacts for the project were achieved through research training of students, a new course offered by PI Sterner and Co-PI Lerman, collaboration on a new dance composition addressing biodiversity loss led by Co-PI Lerman, and contributions to improving the reusability of scientific articles in ecology and evolution. The project supported mentorship and training of 10 undergraduate and graduate students in diverse fields such as philosophy of science, computer science, and sustainability. Sterner and Lerman collaborated on designing and teaching a new course, Creative Tools for Saving Biodiversity, that introduced undergraduate and graduate students in the arts and sciences to techniques for creative thinking and research focused on addressing wicked problems such as global biodiversity loss. Sterner and Co-PI Franz also collaborated with Lerman on a new dance composition she is developing, “Wicked Bodies,” which engages themes related to the extinction of biological species and how different sorts of knowledge figure into a sustainable relationship between people and nature. The composition was originally scheduled to be performed during the project but has been delayed due to the Covid-19 pandemic. A further outcome of project research was to apply a computational method developed for author name disambiguation on large collections of scientific texts in order to correctly identify authors as people across publications where names are duplicated or vary. In a manuscript now under review, the project reports accuracy of over 90% for disambiguating authors on a collection of more than 600,000 scientific articles that have not previously been analyzed. These results will be made open access online and Sterner is pursuing new collaborations to make them available through key repositories such as WikiData.

Last Modified: 11/24/2021
Modified by: Beckett Sterner