Join us for talks, lunch, and catching up with networks researchers at IU
Speakers: Stasa Milojevic, YY Ahn, Sadamori Kojaku
IUNI will provide lunch for all in-person attendees.
Registration is required
Titles and Abstracts
Science of Science. The Science of Science is a research area that aims to provide a deep quantitative understanding of the relational structure between the scientists, institutions, and ideas. It is also interested in the identification of fundamental mechanisms responsible for scientific discovery. In this talk I will provide a brief introduction to this emerging field and showcase a number of studies that use the data from scientific publications to shed light on contemporary research practices (the team work, interdisciplinarity, and productivity) and their effect on the scientific workforce.
Science Genome project: Imagining the space of knowledge with machine learning. In this talk, I will introduce the Science Genome project, which is an ongoing effort to leverage modern machine learning (esp. representation learning) methods to build a unifying, quantitative framework for the Science of Science.
Distilling rich but crude scholarly data using representation learning. Data is the "oil" that fuels our quantitative exploration and exploitation of the main driver of science. However, scholarly data is crude; it is often noisy, imbalanced, and biased. Machine learning algorithms trained on crude data may generate false claims and make biased decisions in high-stake areas such as funding and promotion. In this talk, I will talk about a general embedding framework that can distill rich but crude scholarly data into a compact, useful, and unbiased vector representation. I will demonstrate that the generated vector representation can be directly input to powerful machine learning models, which arguments our ability to quantify, classify, and predict scientific knowledge production.