William Wei Song, Dalarna University, Sweden

He has been general chair, track chair and program committee chair of international conferences workshops, and symposiums, including World Wide Web, ICCIA 2017-2021, and ISD series. He was keynote speaker at ICKET, WAST, ICCIA, etc. He has been reviewer of many scientific and technology foundations, including ITF (Hong Kong), Vinnova (Sweden), EPSRC (UK), and ESPRIT (EEC) and FP6/7 (EU), and NSFC (China). Prof. Song has published over 150 research papers in international journals including Data and Knowledge Engineering (Elsevier), Journal on Computational Logic, and Information Science, and conferences, including Conceptual Modelling (former Entity Relationships Modelling), CAiSE, WWW, WISE, COMPSAC, and ISD. His research interests cover Database Systems, Conceptual Modelling, Web Science, Semantic Web, Computational Social Network, and Big Data analysis (in intelligent transport, smart cities, and e-Healthcare).

(Onsite Talk) Speech Title: A Russellian (formal and semantic) angle to view the issues in the big data analysis field

Abstract: The methods for big data analysis have been booming for the last decade particularly since the introduction of deep learning algorithms. However, when facing the requirements of interpretation of the analysis results, exact reasoning process, and complex representation of domain knowledge, the conventional learning methods may not provide a satisfactory approach and solution to the requirements given above. As said by Bertrand Russell in his book “Introduction to Mathematical Philosophy”, “… the method is more important than the results, from the point of view of further research; and the method cannot well be explained within the framework …”.
In this talk, I intend to address the above-mentioned issues in the big data analysis procedure from the angles of formality, semantics and logics which could lead to different solutions. Considering the power of AI in logic and the semantic web (SW) in semantics, we propose a coordinate-representational framework for the big data (BD) analysis with the concepts of AI, the SW, and BD, where a fundamental representation of data objects is triples, a term borrowed from SW. In this coordinate, the data-knowledge is viewed as a complex object in hyper-structure. In this hyper-structure, an object is viewed, understood, and interpreted in relativity to the e.g. adjacent objects. In relativity to a complex object, a simple object is considered as an object with coarse granularity. Based on the inter-object relationship representation (still on the triples), data reasoning is done in terms of the equality equation theory and knowledge inference is done in terms of the knowledge graphs (KG) acquired from application domains.
As known to all, LLM is a successful application of deep learning methods in NLP (natural language process), particularly in identifying the “meanings” of text through building up text-meaning patterns with ten-millions of training data. And the connections between the sentences may form semantic interference of meanings, which it is suspicious (unclear or uncertain) whether it can compete logic reasoning. In consequence, with an explanation of the above concepts in an example of Large Language Models (LLM), I attempt to offer a formal definition of the key concepts applied in the data analysis processes and structures and aim to pave a novel road (the coordinate-based representation of the relationships among the BG, the AI, and the SW) toward semantic and knowledge-based interferences and analysis, thus leading to a tremendous improvement of big data analysis.

 

 

 

Lianmeng Jiao, Northwestern Polytechnical University, Xi'an, China

Jiao Lianmeng is now an Associate Professor with the School of Automation, Northwestern Polytechnical University, Xi'an, China. He received the B.E. and M.E. degrees from Northwestern Polytechnical University, Xi'an, China, in 2009 and 2012 respectively and the Ph.D. degree from Université de Technologie de Compiègne, Compiègne, France, in 2016. He has authored/co-authored of three books, more than 50 peer-reviewed journal and conference papers. His research interests include information fusion, belief function theory and its application in machine learning and decision making. He has been served as Session Chairs of ICIUS2016/CCC2021/BELIEF2021/PRAI2023, Guest Editor of Entropy and CMES: Computer Modeling in Engineering & Sciences, and editorial board member of several peer reviewed journals including PLOS One and American Journal of Artificial Intelligence. http://teacher.nwpu.edu.cn/jiaolianmeng

(Onsite Talk) Speech Title: Belief rule-based model for interpretable classification of uncertain data

Abstract: Classification is a key tool in data mining/machine learning. It is widely applied in many fields, e.g., military target recognition, medical diagnosis, software defect prediction, intrusion detection, fault diagnosis. However, great uncertainty (incompleteness, imprecision, unreliability, etc.) may exist in practical data classification problems, due to the random nature of collection processes, measurement errors, or insufficient knowledge, etc. These different types of uncertainty bring great challenges to classifier design. Besides, model interpretability is another demanding property for classifier design especially in those security, privacy, and ethical applications. In this talk, I briefly introduce the developed belief rule-based classification (BRBC) model, which can provide interpretable classification for uncertain data. It is developed based on the rule-based classification and the belief function theory. Some variants of the BRBC model, including the compact BRBC for big data, the hybrid BRBCS by integrating data and knowledge, and the dynamic BRBCS for stream data, will also be discussed.

 

Gang Li, Qilu University of Technology(Shandong Academy of Sciences), China

Gang Li, Professor and Doctoral Supervisor of Qilu University of Technology, Young Expert of Shandong Province Taishan Scholar, Winner of the Shandong Provincial May Day Labor Medal，ISO/IEC JTC1 (Information Technology) Registered Expert,ISO/IEC JTC1 (Information Technology) Registered Expert,Member of Standardization Principles and Methods Standardization Technical Committee (SAC/TC 286),Deputy Director of the Interconnection of Information Technology Equipment (TC28/SC25).Engaged in big data analysis, digital economy, digital government and other direction research.In the past five years, he has undertaken more than 40 national key research and development, provincial science and technology major projects,led or participated in the release of 35 national standards and 29 local standards, won 9 provincial scientific and ministerial science and technology awards, published more than 30 SCI / EI index papers, published 5 academic monographs, compiled 1 teaching materials, and authorized 21 invention patents

(Onsite Talk) Speech Title: The Digital Transformation of Industries Driven by Big Data and Large AI Models 

Presented by: Jiachen Li , Qilu University of Technology(Shandong Academy of Sciences), China

Abstract: Large AI models are machine learning models with large-scale parameters and complex computational structures. These models are typically constructed from deep neural networks with billions or even hundreds of billions of parameters, which learn complex patterns and features by training massive amounts of data. They have stronger generalization abilities and can make accurate predictions on unseen data. However, in certain specific tasks or small sample learning contexts, large AI models may not always outperform small models designed for specific domains, especially in industrial applications. How to solve complex problems in the industrial field through large AI models has become an important research topic. This report will explore the characteristics of various large AI models, analyze the problems faced by large AI models in industrial applications, and discuss some works in the construction of high-quality industrial datasets and the development of industrial large AI models. These works will provide reference for promoting the application of large AI models in the industrial field.