Alba (Ruiran) Su
I am currently pursuing a DPhil in Engineering Science at the University of Oxford, specializing in Natural Language Processing (NLP). My research focuses on the intersection of Graph Neural Networks (GNNs) and Large Language Models (LLMs) to address the pressing issue of misinformation, particularly in climate science. My work aims to develop advanced NLP tools to assist online moderators and fact-checkers in identifying and mitigating false claims. In addition to this, I am exploring the potential of knowledge graphs to enhance next-generation telecommunication systems. My research integrates textual and social network data to analyze discussions and identify misinformation spreaders on social media platforms.
Education
- 2022 - Present: DPhil in Engineering Science, University of Oxford.
- 2021 - 2022: Assistant Researcher, Guanghua School of Management, Peking University.
- 2020 - 2021: School of Architecture, Tsinghua University.
- 2019 - 2022: Computer Science, Peking University.
Recent Projects
Intent-based Network Empowered by Knowledge Graph
In recent years, Intent-based Networks (IBNs) have emerged as a revolutionary approach to managing and optimizing complex networks. Nevertheless, modern IBN systems often encounter difficulties in accurately interpreting user intent and translating it into network configurations. Our research introduces an Intent-based Network empowered by Knowledge Graph (IBN-KG), which ingeniously integrates knowledge graph technology with IBN, significantly improving the translation and management of user intents. The study primarily focuses on constructing a knowledge graph for grid scenarios and leveraging this graph to enhance the IBN's performance. Additionally, we present a customized Knowledge Graph construction framework, specifically designed for vertical industry applications, with a special emphasis on smart grid scenarios. This encompasses the development of a unique data layer, extraction of knowledge through natural language processing, knowledge fusion, continuous updating of knowledge, and the application of knowledge through user interfaces. Overall, the fusion of Knowledge Graph technology with IBN heralds a smarter, more adaptable, and efficient method for translating and managing user intent in network configurations, particularly in smart grid applications.
ClimateViz
ClimateViz is a cutting-edge dataset developed to tackle the challenge of misinformation in climate science. In an era where false narratives about climate change can significantly hinder public understanding and policy-making, ClimateViz serves as a foundational resource for the research community. Sourced from reputable organizations such as NOAA, the UK Met Office, and Copernicus, this dataset includes scientific graphics and visual data enriched with factual annotations. ClimateViz is designed to bridge the gap between visual and textual data, enabling researchers to evaluate and fine-tune large vision-language models (LVLMs) for enhanced fact-checking and retrieval-augmented generation (RAG).
Implementing Automatic 5G Network Slicing on the Open Network Automation Platform
Intent-Based Networking (IBN) represents an evolution in networking, wherein network control logic is decoupled, and closed-loop orchestration techniques are employed to automate application intents. An IBN is adaptive and intelligent, capable of interpreting operator intents in real-time to dynamically adjust network configurations and ensure reliability. End-to-end network slicing in 5G and beyond offers immense potential for the telecommunication industry to meet diverse Service Level Agreements (SLAs) for different verticals. However, the realization of this potential requires a smarter and more flexible slicing management framework. Research and standardization bodies such as 3GPP are increasingly focusing on smart slicing, which is geared towards accommodating user intents and requirements through real-time recognition and adaptive network slicing to meet SLAs.
Internship
Publications
Conference Papers:
-
Decoding Climate Disagreement: A Graph Neural Network-Based Approach to Understanding Social Media Dynamics
Ruiran Su, Janet Pierrehumbert (Supervisor),
ACL 2024 | paper -
An Intent-based Network Empowered by Knowledge Graph: Enhancement of Intent Translation and Management Function for Vertical Industry
Dong Wang (Supervisor), Ruiran Su
CIC 2023 | paper -
Trends and Challenges of Policy Verification for Intent-based Networking towards 6G
Dong Wang (Supervisor), Ruiran Su
CIC 2022 | paper -
An Intent-based Smart Slicing Framework for Vertical Industry in B5G Networks
Dong Wang (Supervisor), Ruiran Su
CIC 2021 | paper
International Conference
Awards & Talents
Awards
Talents
