Complexity Science Hub

𝗛𝗼𝘄 𝗳𝗮𝗶𝗿 𝗮𝗿𝗲 𝗔𝗜 𝗲𝘅𝗽𝗲𝗿𝘁 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀? 🤖 𝗜𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝘃𝗲 𝘃𝗶𝘀𝘂𝗮𝗹 𝘀𝗵𝗼𝘄𝘀 𝘄𝗵𝗲𝗿𝗲 𝗟𝗟𝗠𝘀 𝗳𝗮𝗹𝗹 𝘀𝗵𝗼𝗿𝘁. CSH's Lisette Espín-Noboa and colleagues benchmarked various large language models on their ability to recommend academic experts in physics, and how fair these recommendations are ‒ see yesterday's post for details! Together with CSH data visualization expert Yi Zhe Ang, the team published the interactive visualization “𝗪𝗵𝗼𝘀𝗲 𝗻𝗮𝗺𝗲 𝗰𝗼𝗺𝗲𝘀 𝘂𝗽”. 👩💻 Explore how often each physicist was recommended across runs. On the dashbaord and leaderboard, you can view individual chatbot performance per evaluation metric. ℹ️ View the walkthrough video to see how to use the dashboard! 𝗪𝗵𝗮𝘁 𝘁𝗵𝗲 𝗱𝗮𝘀𝗵𝗯𝗼𝗮𝗿𝗱 𝘀𝗵𝗼𝘄𝘀: • 𝟮𝟮 𝗟𝗟𝗠𝘀 from widely used model families, including Gemini, GPT, Llama, Qwen, DeepSeek, Grok, Mistral, and Gemma. • 𝗡𝗶𝗻𝗲 𝗺𝗲𝘁𝗿𝗶𝗰𝘀 covering technical quality and social representation. • Users can 𝗰𝗼𝗺𝗽𝗮𝗿𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 and model families and explore their performance across the different metrics. 𝗦𝗼𝗺𝗲 𝗶𝗻𝘀𝗶𝗴𝗵𝘁𝘀: ✅ 𝗙𝗮𝗰𝘁𝘂𝗮𝗹𝗶𝘁𝘆: LLMs score between 0.63 and 0.82, meaning that recommended physicists correspond to real scientists in the reference database in 63–82% of recommendations. DeepSeek and Gemini score among the highest, while Qwen, Llama, and Mistral score lower. 🌍 𝗗𝗶𝘃𝗲𝗿𝘀𝗶𝘁𝘆: Scores range from 0.44 to 0.69 (i.e., 44–69% of an even distribution across categories like gender or ethnicity), with Deepseek models leading the board. 🎯 𝗣𝗮𝗿𝗶𝘁𝘆: For parity ‒ which measures how closely recommended scholars matched group proportions in the reference database (e.g., men vs. women) ‒ scores ranged from 54–60%, with Gemma models leading. 🚀 𝗥𝗼𝗼𝗺 𝗳𝗼𝗿 𝗶𝗺𝗽𝗿𝗼𝘃𝗲𝗺𝗲𝗻𝘁: “In terms of diversity and parity, there is clearly room for improvement,” says Espín-Noboa. 🔗 𝗘𝘅𝗽𝗹𝗼𝗿𝗲 𝘁𝗵𝗲 𝘃𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝘆𝗼𝘂𝗿𝘀𝗲𝗹𝗳: https://lnkd.in/dUkZ-HVq

𝗧𝗵𝗲 𝟮𝟬𝟮𝟲 𝗩𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗶𝗻𝗴 𝗖𝗼𝗺𝗽𝗹𝗲𝘅𝗶𝘁𝘆 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 is underway! 🚀 This week professionals from a multitude of backgrounds ‒ scientists, journalists, artists, designers, developers, and data analysts ‒ come together to sharpen their skills in the art of visualizing scientific findings. 📊 The workshop, led data by #visualization expert Liuhuaying Yang and information designer Paul Kahn, challenges participants to collaboratively analyze data from current complexity science research problems and then create visualization prototypes in hands-on team sessions. After a short intro, things kicked off with a few rounds of speed dating in the garden. Multidisciplinary teams were formed to foster the exchange of different perspectives during what will be a busy week. CSH researchers Rafael Prieto-Curiel and Katharina Ledebur each presented a current project and datasets for the participants to try and turn into engaging pieces of visual communication. The CSH salon is already buzzing with creative energy as we speak ‒ here's to an exciting and inspiring week! P.S.: Couldn't make it to the workshop but still want to meet the hosts and participants? Sign up for the 𝗗𝗮𝘁𝗮𝗩𝗶𝘇 𝗚𝗮𝗿𝗱𝗲𝗻 𝗣𝗮𝗿𝘁𝘆 𝗼𝗻 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆, 𝟮𝟳 𝗔𝘂𝗴𝘂𝘀𝘁, 7 pm: https://lnkd.in/djMpe5vK P.P.S.: Workshop co-host Paul Kahn just published his book "Mapping Knowledge Across Time" and will present it tomorrow at the Shakespeare & Company book store. More info here: https://lnkd.in/dkAeFkWe

𝗜𝗳 𝘆𝗼𝘂'𝗿𝗲 𝗹𝗼𝗼𝗸𝗶𝗻𝗴 𝗳𝗼𝗿 𝗮𝗻 𝗲𝘅𝗽𝗲𝗿𝘁, 𝘄𝗵𝗲𝗿𝗲 𝗱𝗼 𝘆𝗼𝘂 𝘁𝘂𝗿𝗻? Whether a journalist looking for an interviewee, an event organizer searching for panelists, a decision-maker looking for expert advice, or a student working on a term paper, finding the right expert is part of the process.   Increasingly, AI tools help with that search. But who do they recommend as an expert? And how fair is that, really?   In a trilogy of studies, researchers at the Complexity Science Hub (CSH) took a closer look, benchmarking AI tools on their ability to recommend academic experts. “We tested state-of-the-art LLMs, including open-weight models like Llama and DeepSeek as well as proprietary models like Gemini and GPT,” says lead researcher Lisette Espín-Noboa. 𝗛𝗲𝗿𝗲’𝘀 𝘄𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝗳𝗼𝘂𝗻𝗱:   👨🔬 𝗙𝗶𝗿𝘀𝘁, the researchers examined AI-generated expert recommendations in physics and found that models tended to favor male, senior and highly cited scholars from the US. They also made factual errors in around 20% of recommendations.   📊 𝗡𝗲𝘅𝘁, they introduced a benchmark with nine metrics covering technical quality and social representation, and tested four user interventions across 22 LLMs for recommending physics experts. The interventions improved either factual accuracy or social representation, but not both at the same time.   🌍 𝗙𝗶𝗻𝗮𝗹𝗹𝘆, they examined whether a person’s role, language or geographic location affects the experts recommended across six academic disciplines. Geographic location influenced who was recommended, while language and role did not. Together with CSH data visualization expert Yi Zhe Ang, the team created the interactive #visualization “𝗪𝗵𝗼𝘀𝗲 𝗻𝗮𝗺𝗲 𝗰𝗼𝗺𝗲𝘀 𝘂𝗽”, where users can explore the performance of 22 large language models (LLMs) across nine metrics, including diversity, parity and factual accuracy. 🔗 Learn more: https://lnkd.in/dYNWqvwy 🔗 Explore the visualization: https://lnkd.in/dUkZ-HVq You can find links to the three studies here in the About: https://lnkd.in/dUkZ-HVq. Lisette presented the second study recently at #KDD2026 ACM SIGKDD & Annual KDD Conference. The first and third studies are currently available as preprints.

𝗗𝗮𝘁𝗮 𝗩𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗖𝗿𝗮𝘀𝗵 𝗖𝗼𝘂𝗿𝘀𝗲 at the Complexity Science Hub – just a few spots left! Learn how to #visualize and communicate your data-driven findings. This workshop covers visualization strategies and creative approaches to telling stories with data, led by data visualization expert Liuhuaying Yang (World Data Viz Prize, Information Is Beautiful Award). Explore examples from CSH research: https://csh.ac.at/visuals/ 📅 September 21, 2:00–4:30 pm 📍 Complexity Science Hub, Vienna 👥 Limited to 30 participants (postdocs only; first-come, first-served) 𝗥𝗲𝗴𝗶𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻: https://lnkd.in/dtMJWwTy Note: Registration is only confirmed once you receive a separate confirmation email. If you can't secure a spot, there will be a waiting list. This workshop is part of Postdoc Appreciation Week #PAW Austria (September 21–25). 🤝 PAW Austria is a collaborative national initiative supported by institutions from all over Austria, including: Universität Wien, Technische Universität Wien, Institute of Science and Technology Austria, LBG Career Center, Medical University of Vienna, Universität Innsbruck, Johannes Kepler Universität Linz, Complexity Science Hub, University of Graz, OeAD – Österreichs Agentur für Bildung und Internationalisierung, Technische Universität Graz, University of Music and Performing Arts Vienna, Universität Klagenfurt, Austrian Academy of Sciences - Österreichische Akademie der Wissenschaften (ÖAW), St. Anna Children's Cancer Research Institute (CCRI), Vienna BioCenter, International Institute for Applied Systems Analysis (IIASA), Akademie der Bildenden K��nste Wien.

𝗪𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝘀 𝘁𝗼 𝗮 𝗰𝗶𝘁𝘆 𝘄𝗵𝗲𝗻 𝗽𝗲𝗼𝗽𝗹𝗲 𝗺𝗼𝘃𝗲 — 𝗼𝗿 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗺𝗼𝘃𝗲 — 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗰𝗲𝗻𝘁𝗲𝗿 𝘁𝗼 𝘁𝗵𝗲 𝗼𝘂𝘁𝘀𝗸𝗶𝗿𝘁𝘀? Between 1990 and 2020, Mexico's urban population doubled – and it continues to grow. Yet at the same time, city centers have lost population (around 2.5 million people), according to a study by TECNOLÓGICO DE MONTERREY in collaboration with the Complexity Science Hub, published in #NatureCommunications. The researchers examined 69 cities across Mexico to understand how they're growing. They found that every one of them lost population in its central areas — regardless of the city's size or characteristics. In 1990, more than 40% of the urban population lived in central areas. By 2020, that share had fallen to 22%. 🚗 𝗪𝗵𝗲𝗻 𝗰𝗶𝘁𝗶𝗲𝘀 𝗴𝗿𝗼𝘄 𝗼𝘂𝘁𝘄𝗮𝗿𝗱 When residents leave city centers and cities sprawl outward, the consequences can extend across urban life — including longer commutes, poorer access to public services such as hospitals and government offices, and greater reliance on cars. These changes reshape how people live, move and access the city — with consequences for quality of life, the environment, and much more This study is a collaboration between TECNOLÓGICO DE MONTERREY (Centro para el Futuro de las Ciudades) and the Complexity Science Hub, involving Gonzalo G. Peraza Mues, Eugen Resendiz, Rodolfo Figueroa Soriano, Rafael Prieto-Curiel, and Roberto Ponce Lopez. 🔗 Read the study: https://lnkd.in/d7Wyzacr 🇦🇹 𝗣𝗦: 𝗔 𝗴𝗹𝗶𝗺𝗽𝘀𝗲 𝗼𝗳 𝗩𝗶𝗲𝗻𝗻𝗮 Separate from the study, Rafael Prieto-Curiel also looked at how Vienna's population has shifted over time. In 1950, nearly 600,000 people – about 37% of the city – lived in the inner districts (1–9). By 2000, that had dropped to 35%. Since then, the inner districts have gained population back. But the outer districts are growing much faster.

𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗮𝗿𝗲 𝗼𝗽𝗲𝗻 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗚𝗿𝗮𝗱𝘂𝗮𝘁𝗲 𝗣𝗿𝗼𝗴𝗿𝗮𝗺 𝗶𝗻 𝗖𝗼𝗺𝗽𝗹𝗲𝘅𝗶𝘁𝘆 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 at the Complexity Science Hub! This year’s call is open until October 15, 2026. ❓ 𝗛𝗼𝘄 𝗱𝗼 𝘆𝗼𝘂 𝗸𝗻𝗼𝘄 if the program is right for you ❓ If you want to use science to help change the world and create real impact. If you’re excited about combining large datasets with quantitative modeling and rigorous theory. If you’re driven to tackle critical real-world challenges. And, a little more technically: if you have a background in the natural or social sciences or mathematics, with a strong foundation in quantitative approaches. 📅 Apply by 15 October 2026 👉 More information: https://lnkd.in/dEGntxp9 ___ Graduate students at CSH will benefit from the collective expertise of our members, gaining valuable insights and mentorship as they build the foundation for successful academic and professional careers. Members of the Complexity Science Hub are AIT Austrian Institute of Technology, University of Natural Resources and Life Sciences, Vienna (BOKU), Central European University, Interdisciplinary Transformation University (IT:U), Medical University of Vienna, Technische Universität Wien, Technische Universität Graz, Universität für Weiterbildung Krems, Vetmeduni, WU (Vienna University of Economics and Business) and Austrian Federal Economic Chamber. #PhDOpportunities #ComplexityScience #DataScience #NetworkScience #Research #ComplexityScienceHub

I feel so lucky to call Vienna my home and to have met so many inspiring people here. 😊 I’m grateful for all the opportunities I’ve had here and, of course, truly humbled to have received this award as well. 🤗 https://lnkd.in/dKJbwcJU

🏆 𝗛𝘂𝗴𝗲 – 𝘄𝗲𝗹𝗹, 𝗱𝗼𝘂𝗯𝗹𝗲 𝗵𝘂𝗴𝗲 – 𝗰𝗼𝗻𝗴𝗿𝗮𝘁𝘂𝗹𝗮𝘁𝗶𝗼𝗻𝘀 to Elma Hot Dervic and Jan Korbel, who have both been awarded a #Förderungspreis of the City of Vienna 2026! 🎉 The two Complexity Science Hub researchers are looking into questions that may at first seem worlds apart. Yet underneath, they share a common thread: understanding how patterns emerge from the interactions of countless individual elements. And now they share something else – the City of Vienna’s Förderungspreis 2026, awarded for outstanding scientific work. 𝗝𝗮𝗻 𝗞𝗼𝗿𝗯𝗲𝗹 explores the #foundations of complex systems, using tools from modern physics to study societal dynamics. His work on #polarization, collective behavior, and information processing bridges the social sciences and complexity research, offering important contributions to our understanding of complex social systems. 𝗘𝗹𝗺𝗮 𝗗𝗲𝗿𝘃𝗶𝗰 works at the intersection of #medicine and data science. Drawing on large-scale medical datasets comprising millions of data points, her research identifies patterns associated with disease risks and inequalities in healthcare. Congratulations, Elma and Jan – very well deserved! 👏👏👏 🔗 More: https://lnkd.in/dyJe3eia ___ The City of Vienna’s “Förderungspreise” honor individuals whose work opens up new perspectives and helps shape Vienna’s cultural, scientific and educational life. The awards are presented by City of Vienna’s Mayor Michael Ludwig and City Councillor for Culture and Science Veronica Kaup-Hasler. 🔗 More: https://lnkd.in/eYu3gx_C

🌟 #Jobalert 🏥 Are you passionate about turning health data into insights that can make a difference? The Complexity Science Hub is looking for a #postdoctoral researcher to join the TRACK-PEM project, led by Katharina Ledebur. 𝗪𝗵𝗮𝘁 𝗶𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 𝗮𝗯𝗼𝘂𝘁? The project investigates post-exertional malaise (#PEM) in people with 𝗠𝗘/𝗖𝗙𝗦 (Myalgic Encephalomyelitis/Chronic Fatigue Syndrome) and post-COVID condition and aims to find out whether patterns in wearable data can help identify and classify PEM and related changes in health. 𝗪𝗵𝘆? PEM is currently assessed mainly through interviews and questionnaires. Objective measures that capture how it develops in everyday life are still lacking. 𝗪𝗵𝗮𝘁 𝘄𝗼𝘂𝗹𝗱 𝘆𝗼𝘂 𝗱𝗼? This position involves methodological development and analysis of complex time-series data, with a particular focus on individual variability and real-world health data. If you want to join a research group dedicated to understanding healthcare systems as complex adaptive systems, and work closely with researchers across computational health, network medicine, epidemiology, data science, and clinical research – apply now. Learn more about 𝗛𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲 & 𝗠𝗲𝗱𝗶𝗰𝗶𝗻𝗲 at the Complexity Science Hub: https://lnkd.in/ex2fBnG2 🕐 2-year fully funded postdoc 📅 Ideal start: December 2026 👀 Applications are reviewed on a rolling basis 👉 Learn more: https://lnkd.in/dmuJMmxT

No. 2️⃣ Research Topic of the month: 𝗗𝗮𝘁𝗮 𝗩𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 – meet Liuhuaying Yang! How does she turn data into interactive, explorable visualizations? What is the process behind it? And why does she think it is so important to make data accessible in this way? In this interview series, our researchers give you a behind-the-scenes look into their field. Here you can find all CSH #visualizations: https://csh.ac.at/visuals/ ––– 𝗔𝗯𝗼𝘂𝘁 𝘁𝗵𝗶𝘀 𝘀𝗲𝗿𝗶𝗲𝘀: In June 2026, the Complexity Science Hub celebrated its 10th anniversary — and throughout our anniversary year, we're spotlighting one research topic every month. Curious about how we celebrated? Find our photos here: https://lnkd.in/dWixmffH — and for a deep dive into a decade of complexity science, read our 10-year publication here: https://lnkd.in/e7s_CZHv #ComplexityScience #10YearsCSH #CSHresearch #ResearchTopicOfTheMonth #ComplexSystems