Computational Social Science and Complex Systems
For the last two decades, computational models combined with large-scale transactional data have been used by computational social scientists to explain, predict, and even forecast social phenomena. The general aim of this special session is to address social phenomena emerging over multiple scales, ranging from the interactions of individuals to the emergence of self-organized global movements within the realm of computational social science. We would like to gather researchers from different disciplines and methodological backgrounds to form a forum to discuss ideas, research questions, recent results, and future challenges in this emerging area of research and public interest.
Data Science for Social and Behavioral Analytics (DSSBA 2022)
In this era where mobile devices and online services occupy a major role in daily lives, more and more data are recorded about the behavior of humans and their social interactions. For instance, data are collected about interactions between social network users, and between learners using e-learning systems, about the shopping behavior of customers, and about medical pathways in hospitals.
To make sense of behavioral and social data, Behavior Analytics (BA) has emerged as a key research area in data science. The aim is to analyze data to gain a better understanding of behavior, which can support taking better decisions, but also to design machine learning models to offer tailored services to users such as personalized recommendation. Analyzing behavioral and social data raises several challenges such as (1) designing appropriate, scalable and efficient algorithms and models for analyzing behavioral data, (2) preserving the privacy of users for behavioral analytics, (3) analyzing behavior by taking into account the cognitive and social dimensions, (4) designing intelligent systems and services that are powered by behavioral and social data models, and (5) addressing the privacy and security issues in algorithms, models, and tools for social and behavioral analytics.
This special session aims at bringing together the Data Science, Machine Learning, Business and Cognitive Science communities to present advances on the design of efficient, scalable and effective solutions for analyzing social and behavioral data.
Data Science and Advanced Analytics for Autonomous Vehicles
Autonomous vehicles (AVs), i.e., self-driving cars and ships, have attracted great attention worldwide in recent years. Especially, during the epidemic of COVID-19, the expectation for autonomous vehicles is more urgent because the AVs would be capable of transferring a huge number of patients and medical resources, reducing the risk of infection among passengers and drivers, and minimizing the risk of virus transmission through transportation networks. To be capable of making optimal decisions, AVs should have the ability to perceive complex dynamic environments quickly and accurately. Fueled by big data from various sensors and advanced computing resources, data science and advanced analytics (DSAA) have become an essential component of AVs and stimulated the development of AVs.
This special session will bring together researchers, industry experts, and practitioners who are interested in cultivating specialized and important aspects of data science and analytics in the context of autonomous vehicles
Practical applications of explainable artificial intelligence methods (PRAXAI)
This special session focuses on bringing the research on Explainable Artificial Intelligence (XAI) to actual applications and tools that help to better integrate them as a must-have step in every AI pipeline. We welcome papers that showcase how XAI has been successfully applied in real-world AI-based tasks, helping domain experts understand the results of a model. Moreover, we also encourage the submission of novel techniques to augment and visualize the information contained in the model explanations. Furthermore, we expect a presentation of practical development tools that make it easier for AI practitioners to integrate XAI methods into their daily work.
Data Science for Cyber and National Security (DS4S)
National security protects the country against substantial physical and psychological threats to our government; public safety; environment; or energy, food, and fiscal infrastructures. Terrorism, misinformation, and cyberattacks are common examples which are among the top 26 national security threats America is currently facing. Data-driven security is an emerging interdisciplinary area that focuses on researching and applying data science to solve national security problems. For instance, it deals with applying social network analysis and game theory for bad actor detection and counter-attack in crime, terrorist, and nuclear proliferation networks; using data science to reduce the spread of misinformation which is responsible for manipulating opinions and public response; integrating data science and analytics into cybersecurity for meeting the cybersecurity challenges of processing large data sets in order to gain valuable insights and reduce cybersecurity risks.
Data Science and Artificial Intelligence Enabled Trustworthy Recommendations
Nowadays, the renaissance of artificial intelligence (AI) has attracted huge attention in everyday real life. Recommender systems, as one of the most popular applications of AI, have already become an indispensable means for helping web users identify the most relevant information/services in the era of information overload. The applications of such systems are multi-faceted, including targeted advertising, intelligent financial assistant, and e-commerce, and are bringing immense convenience to people’s daily lives. However, despite rapid advances in recommendation, the increasing public awareness of the trustworthiness of recommender systems has now introduced higher expectations on relevant research.
The aim of this special session is to engage with active researchers from recommendation communities and deliver the state-of-the-art research insights into the core challenges in the algorithmic trustworthiness. Firstly, the unprecedentedly growing heterogeneity in real-world recommendation data has been challenging the adaptivity of contemporary algorithms to various settings, e.g., interest drift of users, cold start users/items, highly interaction sparsity, and multimodal content. Secondly, trustworthy recommendation approaches should also be robust, secure, interpretable, privacy-preserving, and fair. Consequently, trustworthiness is becoming an important performance indicator for state-of-the-art recommendation models in addition to accuracy. In light of these emerging challenges that co-exist with previous techniques and applications for recommendation, this special session focuses on novel research in this field with the notion of trustworthiness. The special session will provide an opportunity to promote the research on trustworthy recommender systems, thus developing beneficial AI applications and better universalizing the advanced techniques to a wider range of the real life.
PSTDA2022 - Special Session on Private, Secure, and Trust Data Analytics
The fusion of scalable computing infrastructure, big data, and artificial intelligence has boosted the development and application of data science and advanced data analytics. However, the recently emerging threats on the privacy, security, and trust (PST) of the data and the analytics models have shown a dramatically increasing trend with the wide deployment of data analytics applications. Specifically, the PST attacks on data or models such as model inversion attacks, membership inference attacks, data poisoning attacks, evasion attacks, and model backdoors, have severely made advanced data analytics highly vulnerable, particularly in common scenarios where data are distributed or computation is outsourced like MLaaS (Machine Learning as a Service). On the other hand, defence solutions are proposed as new computing schemes, PST frameworks, algorithms, and methods. For example, differential privacy, federated learning, and machine unlearning are proposed for privacy protection in data analytics, and adversarial machine learning is proposed to achieve robust, secure, and trustworthy data analytics. Given the importance and urgence, this special issue aims to provide a venue for researchers, practitioners and developers from different background areas relevant to PST and data analytics to exchange their latest experience, research ideas, and synergic research and development on fundamental issues and applications about privacy, security, and trust issues in data analytics, as a strong supplement to the main train of data science and advanced analytics.
This special session mainly focuses on the discussions of privacy, security, and trust in data analytics, which generally covers (but not limited to) the topics in privacypreserving technology, privacy attacks, federated learning, machine unlearning, data poisoning attacks, model evasion attacks, adversarial learning, model robustness, secure machine learning integrating cryptographic techniques, blockchain techniques protection PST of data and models, etc. This special session invites authors to submit original research work that demonstrate and explore current advances in all related areas mentioned above.