The World Internet Conference 2023, held in Wuzhen, Zhejiang Province, announced its awards for Pioneering Science and Technology on November 8. Among 15 internationally representative projects that received the award, two were from Tsinghua University's Department of Computer Science and Technology. The team led by Professor Sun Maosong was awarded for their "Multi-Level Multi-Type Knowledge Fused Deep Learning Fundamental Methods for Natural Languages" research project, and the team led by Professor Li Guoliang was awarded for their "GaussDB: Distributed Database" research project.
The project "Multi-Level Multi-Type Knowledge Fused Deep Learning Fundamental Methods for Natural Languages" aims to address the three challenges of natural language processing (NLP) in the context of deep learning – the insufficient ability of word-level embedding representations to utilize linguistic knowledge comprehensively, the inadequate proactive awareness of world knowledge by sentence-level language models, and the weak ability of language models to utilize knowledge systems. The project studies and proposes multi-level multi-type knowledge fused deep learning fundamental methods for natural languages in a quite systematic way, overcomes several bottlenecks that restrict the advancement of deep learning in NLP, improving the basic capabilities of NLP models. Based on a number of papers published in top international conferences, the project has released the related open-source tools on GitHub, producing an active influence in the domain of deep learning for natural language integrated with a variety of knowledge. The resulting technology has been successfully implemented in Huawei Cloud, serving over 150 countries and regions with notable effects.
The project “GaussDB: Distributed Database” is designed to meet the database challenges in terms of high performance of massive data, high availability of database services, elasticity and scalability, intelligent optimization of a large scale of database instances, and anti-tampering protection to ensure data security. The project has achieved breakthroughs in various aspects of distributed database technology, including high performance breakthrough in distributed query optimization and transaction processing technology, achieving leading performance in the industry; high availability breakthrough in multi-level disaster recovery technology with fault self-awareness, building China’s first co-located dual-cluster “RPO=0” solution, implementing multi-level high availability such as node level, active zone level and regional level, etc.; high elasticity breakthrough in separation of computation, shared memory, and storage as well as elastic scaling, implementing senseless elastic scaling of instances; high intelligence breakthrough in AI-powered database optimization technology to achieve autonomous database optimization; and, high security breakthrough in multi-faceted security technologies to build full confidentiality and tamper-proof data processing capabilities for protecting the security and privacy of sensitive data. GaussDB's achievements have been widely applied by financial and governmental enterprises in national key infrastructure.
The World Internet Conference Awards for Pioneering Science and Technology targets at rewarding the internationally advanced Internet scientific and technological achievements, advocating international communication and cooperation in this field. The award intends to build an innovative exchange and display platform for leading Internet achievements with high standards, direct the future development of Internet technology, and open up a broad space for international exchanges, mutual learning and the transformation of scientific and technological achievements.