Science Studies lab seminar

Please join us for an open meeting of the Science Studies Laboratory on Friday, April 5th, from 1 to 2.30pm.

Dr Artem Chumachenko will give a talk “Thermodynamic Framework for the Knowledge Network Analysis”, followed by a discussion. 

The meeting will be held in English.

Please find below the abstract of the talk and the speaker’s bio.

The meeting will be held hybrid: in EUROREG (room 1) and online on Zoom: https://uw-edu-pl.zoom.us/j/93795537319?pwd=Y0xkckhndFdzbWl3ajhCcSt4SU5qZz09.

Abstract

Understanding information processing in complex networks is a central focus across various scientific domains. Statistical physics aims to uncover the underlying patterns and structures governing physical systems, while information theory deals with extracting and transmitting valuable information from noisy datasets. Neuroscience endeavors to model the evolution of intricate biological systems, and social sciences strive to discern patterns in diverse social interactions. The rapidly evolving field of machine learning seeks to refine network training models and elucidate the transformation of information into knowledge.

Just as thermodynamics was pivotal in the Industrial Revolution, infodynamics is emerging as a cornerstone of the modern digital era. Statistical mechanics and thermodynamics can be viewed as special cases of information theory, offering profound insights into information processing in complex systems. Applying statistical mechanics methods to the challenges of information processing in complex networks promises fresh perspectives on their state properties and evolutionary principles.

In this presentation, we propose a thermodynamic framework for analyzing complex knowledge networks built from scientific concepts extracted from scholarly documents. Focusing on a corpus spanning over a decade of research in High-energy physics, we employ mono-node approximation to construct thermodynamic knowledge network state diagrams. Our analysis delves into the dynamics of thermodynamic potentials associated with concepts and the role of entropy production. We introduce the notion of network dynamic complexity as a means of quantifying the amount of valuable information, akin to thermodynamic work.

Speaker

Artem Chumachenko is an assistant professor of Science Studies Laboratory at the Centre for European Regional and Local Studies (EUROREG), University of Warsaw. He holds a PhD in Theoretical Physics (National Taras Shevchenko University of Kyiv, Ukraine 2008). As a visiting researcher at Physikalisch-Technische Bundesanstalt (Germany) and International Center for Theoretical Physics (Italy) he specialized in condensed matter physics, theory of superfluidity and quantum computations. Artem's present scientific interests are related to scientometrics, theory of information and machine learning. His work in particular was supported by Swiss National Science Foundation Grant “Complex Information Network Manipulation” and conducted in École Polytechnique Fédérale de Lausanne, Switzerland.