Modeling Multi-Platform Information Diffusion in Social Media: Data-Driven Observations

Accurately modeling information diffusion within and across social media platforms has many practical applications, such as estimating the size of the audience exposed to a particular narrative or developing intervention techniques for addressing misinformation. At the same time, advances in machine learning techniques and the availability of large collections of social media datasets promise significant advances in understanding how to model user activities and user interactions on these platforms. Yet real data reveal phenomena that pose significant challenges to modeling: events in the physical world affect in varying ways conversations on different social media platforms; coordinated influence campaigns may swing discussions in unexpected directions; a platform’s algorithms direct who sees which message.  This key talk discussed the challenges encountered in modeling social media activity in various contexts, from announcements of software vulnerabilities to online reactions during political crises. 

You can view the presentation here.

_{Prof. Dr. Adriana Iamnitchi is Professor of Computer Science and Engineering at the University of South Florida. She holds a PhD in Computer Science from The University of Chicago. Her research lies at the intersection of distributed systems, social computing, empirical analysis of phenomena in online social environments and designing solutions for modeling them.} 

Does explainability imply shifted responsibility?

Developments in domains like Artificial Intelligence, Internet of Things and Data Science lead to possibilities for more autonomous systems that directly interact with their environment. However, this poses a range of questions about responsibility whose answers need to supply explicit directions for security constraints, privacy regulations and traceability demands. This technical talk discussed how the allocation of responsibility by autonomous information systems, system designers, system users, and data subjects is affected by explainability and understandability. In the case of a black-box decision making system, responsibility is allocated more at the designer or system side. In case of a white box decision making system, the responsibility may be shifted to the user and client side. 

You can view the presentation here.

_{Prof. Dr. Anna Wilbik is a professor in Data Fusion and Intelligent Interaction at the Department of Data Science and Knowledge Engineering (DKE) at Maastricht University. She holds her PhD in Computer Science from the Systems Research Institute at the Polish Academy of Science in Warsaw. Currently she is a also vice-chair of the Fuzzy Systems Technical Committee (FSTC) and chair of task force on Explainable Machine Learning of CIS IEEE.}

Reporting robust research results – perspectives from data science in the biomedical field

Responsible data science within scientific research needs to encompass all aspects of the scientific pipeline, from data acquisition through to the presentation and publishing of results and beyond.  This technical talk focused on how we can responsibly present our research to the scientific community, thinking about the obvious and not-so-obvious steps we can take to avoid the presentation of false positive results.  The examples were taken from the biomedical domain, but the principles apply to other domains, and even to the presentation of results outside of the academic sphere.  

You can view the presentation here.

_{Dr. Rachel Cavill is an Assistant Professor in systems biology and applied mathematics within the Department of Data Science and Knowledge Engineering (DKE). After receiving her MMATH in Mathematics and Computer Science from the University of York in 2002, she continued at the same institution and received her PhD in 2007, developing bio-inspired machine learning algorithms. During her postdoc years, she transitioned into the field of bioinformatics and spent four years working at Imperial College London. Currently, she and her team explore how machine learning and data science approaches can allow us to integrate biological datasets from different sources.}

AI in public policy: opportunities and challenges

The use of artificial intelligence (AI) in the public sector has matured over the last few years. The number of AI applications has increased, as have their effectiveness and impact, as applications move beyond pilots and prototypes into real world use in policy making. However, the use of AI in governmental processes is not without its risks, as demonstrated by a SyRI court ruling that the system is not compliant with human rights legislation. While the use of AI can be useful for making predictions or automating decision making, challenges remain regarding the availability of datasets, bias in data and the transparency and accountability of such decisions.

Therefore, it is useful to first undertake decision-making experiments in order to understand downstream opportunities and risks. Such experiments may also elicit how the use of these technologies impact public values, such as the requirements for AI formulated by the High-Level Expert Group on AI. Specific opportunities and challenges of the use of AI in the public sector were presented as well as the role of experiments to seizing opportunities and mitigate risks.

You can view the presentation here.

_{Dr. Anne Fleur van Veenstra is a Senior Scientist and the Director of Science Strategic Analysis and Policy at TNO. As a senior researcher, she has over fifteen years of experience in the fields of Digital Government, Public Administration and Information Systems. She is widely published on data-driven policy making, the societal impact of big and open data, emerging technologies such as artificial intelligence in public service delivery, and the impact of digitalization on organizations and governance. She also set up TNO's Policy Lab initiative, conducting experiments with data-driven policy making in collaboration with government organizations.}