News
10/2023 Workshop - Modeling Knowledge and Evolution of Science
Last month, from 10th to 13th of October, the ModelSEN project welcomed almost 30 collaborating scholars to a 4-day workshop at the Max-Planck Institute for the History of Science in Berlin. The workshop consisted of five sessions dedicated to a different methodological perspective and their application for the modeling of knowledge and the evolution of science. Lightning talks and focused discussions were held on the topics of Opinion Dynamics & Social Networks, Agent-based Modeling, Time Series Analysis & Complex Systems, Language Modeling, and Ontologies. In each of those sessions, participants presented their research and discussed how each approach could contribute to understanding the dynamics of knowledge systems. The workshop yielded very interesting and fruitful discussions about both common ground and differences in the various digital approaches that were presented, bringing together the participant's wide variety of perspectives and disciplinary backgrounds ranging from complexity science, mathematics, physics, biology, psychology, and chemistry, to history, philology, philosophy, sociology, digital humanities, sustainability sciences, information science, linguistics, and archeology.
Building on the workshop's results and central discussion points, the workshop participants and the ModelSEN-team will further explore the use of these diverse digital methods and approaches for the history of science and knowledge systems in future collaborative work.
04/2023 ModelSEN@KSW: Digitally tracing Goethe's Weimar
End of April, ModelSEN teamed up with the Klassikstiftung Weimar (KSW) to dig into the digital traces Goethe and his social network left during the Weimar Classicism. In a two and a half day hackathon, the teams goal was to get a first glimpse on the potential of the newly digitized data of the KSW by applying methods commonly used by the ModelSEN team.
A first group deep-dived into Goethe's Begegnungen und Gespräche (BuG), a collection of biographical sources on Goethe's life that entail encounters (in a wide sense) of Goethe with his contemporaries. Here, a team was able to present a Jupyter Notebook of descriptive statistical analysis on the places and persons mentioned in the BuG over time as well as a first iteration of a social network visualizing Goethe's mental and social network. Another team focused on the letters received and sent by Goethe. Here, it was possible to build ego networks of different periods of Goethe's correspondence and graph-based visualizations of the relevance of senders and receivers. Preliminary results show the particular relevance of the poet Friedrich Schiller and Goethe's ministerial colleague Christian Gottlob von Voigt.
In conclusion, the hackathon was successful — regarding the content as well as the team-building level and we look forward to continuing the cooperation with the KSW.
04/2023 Workshop: Agent-based Modeling at Digital History 2023 (HU Berlin)
We will hold a workshop on 'Growing and pruning the Republic of Letters: Learning to simulate the past with agent-based modeling using the mesa package' at the Digital History Tagung 2023 in Berlin, hosted by the Humboldt-University. We warmly invite anyone interested in simulation methods for historical inquiry to participate! On: Tuesday, 24th May, 9am-1pm (with breaks). Apart from basic python familiarity no previous knowledge is required, and researchers of all backgrounds and levels are welcome.
04/2023 OpenAccess: Agent-based Model on socio-epistemic processes of knowledge spread published
Bernardo Buarque published a NetLogo ABM model on the sharing-network CoMSES.net. Based on principles of opinion dynamics and using MPG-data on the Renaissance of the General Theory of Relativity in the model researcher-agents move over a lattice representing mental models while interacting in their social network. The model allows to examine the connection between agents's accumulated knowledge, social learning, and the span of attitudes towards mental models in an artificial society, thereby offering a quantitative and inductive way to research on the evolution of science.
03/2023 Jupyterbook: GPT-3 Abstract Classification for Scientometrics
A new JupyterBook is out in the Compendium. Jochen Büttner used his expertise in ML to build a case study on how to use ChatGPT3 to classify scientific disciplines. The classification reaches very good results after some finetuning.
12/2022 First release of Scientific Communication Models
As a contribution to the SkillNet conference in Utrecht, we released a beta version of the Scientific Communication models package showcasting the idea of global agents exchanging letters and in that process adopting new ideas. This software package will collect the various agent-based models of the ModelSEN project.
11/2022 Jupyterbook: The formation of the field of Social Network Analysis
A new JupyterBook is out in the Compendium (in German). In the framework of the TU seminar, Lea Weiß and Laura von Welczeck looked at the role of physicists in the transformation of the field of Social Network Analysis and analyzed the information flows and interdisciplinarity in this process using network theoretic approaches.
10/2022 Webinar: Using Dimensions for historical research: Biases, pitfalls and a spark of hope
The use of bibliometrics resources for historical research poses different challenges than those we see commonly in their analyses for the evaluation of researchers and institutions. Some main challenges are for example the completeness of these archives, the validity of qualifiers or the coverage of different languages. These questions are hard to answer, but in this Dimensions webinar we talk through an alternative perspective and showcase an approach to these questions that at least starts the journey to a solution.
10/2022 Whitepaper published
A joint whitepaper of many attendees of the ModelSEN workshop in April 2022 has been published on Zenodo. It brings together the desiderata and future goals of people working with HNR methods, with a focus on the challenges encounterd in the field of history of science. Access the whitepaper here.
09/2022 Survey: Best-Practices and Tools in Historical Network Research
Addressing scholars working on the analysis of historical networks, this survey asks about views on common tools, practices and challenges with the explicit goal to inform the development of future research software and educational resources. Please, share your opinions and experiences in this survey and feel free to reshare the LINK !
08/2022 Software package for 3D citation networks
A new software package is out! Together with Robert Egel we developed a visualization interface for Dimensions Analytics publication data, where for a source paper links to its references and citations as well as the references of references and citations of citations form the basis for the papers 'scientific tree'. This tree is visualized in 3D in your browser and takes into account each papers 'Field of Research'. Have a look at the requirements for using it here: CitationNet documentation. The source code is available on our project Gitlab.
07/2022 Collaboration partners added
After the successfull workshop in the beginning of April a list of partners has now been added to the website. Have a look at the various topics that are of interest here: Cooperation partners
02/2022 Cooperation with project NeWorld@a
ModelSEN will be cooperating with the EU H2020 project NeWorld@a with a focus on science diplomacy.
12/2021 Software package for semantic layer released
The software package semanticlayertools will be collecting approaches to model different aspects of the socio-epistemic layer creation. In the first release, two main approaches are already included. Using the concept of temporal clusters the software allows to find clusters on the semiotic layer of articles citing each other, or being cocited together, and reports how found clusters in time can be characterized on the semantic level by employing topic modelling tools. Additionally, groups of most active researchers or institutions are reported, which allows historical experts to judge the quality of clustering. Based on a scoring algorithm for in a corpus occuring ngrams, the second approach links publications in temporal slices via co-authorship, weighted title or abstract ngrams into a multilayer network. Using a random walker approach multi-node clusters are found, which allow to visualize the flow of importance across time. The full source code is made available on the projects Gitlab.
07/2021 Welcome to our new team member Bernado
The team for our project is finaly complete: Bernardo Buarque is joining the team as an expert in innovation research. With his focus on economic history, he is working on implementing agent-based modelling of research processes.
06/2021 Science communications: Exploring the Expedition for Knowledge with the DFG
As a part of the 100th anniversary of the German Research Foundation, an experimental theater project was initializied to bring the questions of the general population in contact with scientific expertise. The project had to be postponed due to COVID (see https://dfg2020.de/unsere-expedition/) but could finally take place in a reduced form in Berlin. In two different formats scientists from accross Germany came into contact with the interested public. In Blind Dates for Reseach people where asked to guess the research questions Malte Vogl is following in the ModelSEN project, only guided by three objects related to it. The event was recorded and will be published as a video documentation.
06/2021 Research Data Management
The guideline for the research data managment of the ModelSEN project are out in version 1. As we know, software changes and new modelling brings new data. So if in the course of the project practices for data change, we will update this document.
05/2021 New ADHO SIG approved
A part of the outreach of the project deals with the technical side of modelling socio-epistemic networks. In the course of the last four years a grassroot group of tech-oriented people in DH build up the DHTech group, which is now approved as a special interest group (SIG) by the the Alliance of Digital Humanities Organizations (ADHO). Malte Vogl, together with Julia Damerow, will be the convener of the SIG.
04/2021 Seminars at the TU Berlin
This summer we are giving two project-related seminars at the TU Berlin, in the History of Science department. Dirk and Malte are co-teaching the use of 'Socio-epistemic networks for the qualitative and quantitative Description of Knowledge Systems in the History of Science', while Roberto is introducing the 'Histories and historiographies of the relativity revolution'.