Publications

Zero-Shot Relational Learning on Temporal Knowledge Graphs with Large Language Models

The first to draw attention to zero-shot relational learning in the context of temporal knowledge graph forecasting. Design an LLM-empowered method to enhance current embedding-based state-of-the-art models in the zero-shot setting.

Zifeng Ding* & Heling Cai*, Jingpei Wu, Yunpu Ma, Ruotong Liao, Bo Xiong, Volker Tresp

2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics (preprint version)

Oral presentation.

Text2Loc: 3D Point Cloud Localization from Natural Language

Introduce Text2Loc, that fully interprets the semantic relationship between points and text for 3D point cloud locolization. Propose a matching free fine localization method to refine the location predictions, which completely removes the need for complicated text-instance matching. The proposed fine localization method is lighter, faster, and more accurate than previous methods.

Yan Xia* & Letian Shi*, Zifeng Ding, João F Henriques, Daniel Cremers

The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024 (preprint version)

ForecastTKGQuestions: A Benchmark for Temporal Question Answering and Forecasting over Temporal Knowledge Graphs

Propose a new task called forecasting TKGQA focusing on forecasting question answering over temporal knowledge graphs. Also propose the first forecasting TKGQA benchmark ForecastTKGQuestions coupled with a model ForecastTKGQA.

Zifeng Ding* & Zongyue Li* & Ruoxia Qi*, Jingpei Wu, Bailan He, Yunpu Ma, Zhao Meng, Shuo Chen, Ruotong Liao, Zhen Han, Volker Tresp

22nd International Semantic Web Conference (2023) (preprint version)

Improving Few-Shot Inductive Learning on Temporal Knowledge Graphs using Confidence-Augmented Reinforcement Learning

Design a confidence-aware reinforcement learning method to improve few-shot inductive learning on the sparse entities in temporal knowledge graphs. Design a concept regularizer to guide the reinforcement learning process by leveraging the concept information stored in the knowledge bases.

Zifeng Ding* & Jingpei Wu*, Zongyue Li, Yunpu Ma, Volker Tresp

European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (2023) (preprint version)

ECOLA: Enhanced Temporal Knowledge Embeddings with Contextualized Language Representations

The first to enhance temporal knowledge embedding with temporally relevant textual information while preserving the time-evolving properties of entity embedding. Construct three datasets to train the text-enhanced tKG models. Extensive experiments show that ECOLA significantly enhances temporal KG embedding models with up to 287% relative improvements regarding Hits@1 on the link prediction task.

Zhen Han* & Ruotong Liao*, Jindong Gu, Yao Zhang, Zifeng Ding, Yujia Gu, Heinz Koeppl, Hinrich Schuetze, Volker Tresp

Findings of the Association for Computational Linguistics (2023)

Few-Shot Inductive Learning on Temporal Knowledge Graphs using Concept-Aware Information

The first to study reasoning over few-shot sparse entities in temporal knowledge graphs (TKGs) . Propose three benchmark datasets for the new task TKG few-shot out-of-graph link prediction. Propose a model for solving the new task and mine the concept information from the underlying knowledge base for performance enhancement in the few-shot setting.

Zifeng Ding* & Jingpei Wu*, Bailan He, Yunpu Ma, Zhen Han, Volker Tresp

Automated Knowledge Base Construction (2022)

Oral presentation. Honorable mention.

Learning Neural Ordinary Equations for Forecasting Future Links on Temporal Knowledge Graphs

Use neural ordinary differential equations to model the temporal dependencies among the graph information from different historical timestamps for future forecasting.

Zhen Han* & Zifeng Ding*, Yunpu Ma, Yujia Gu, Volker Tresp

Conference on Empirical Methods in Natural Language Processing (2021)