EMNLP Findings2023
TabPrompt: Graph-based Pre-training and Prompting for Few-shot Table Understanding
Rihui Jin, Jianan Wang, Wei Tan, Yongrui Chen, Guilin Qi, Wang Hao
Abstract
Table Understanding (TU) is a crucial aspect of information extraction that enables machines to comprehend the semantics behind tabular data. However, existing methods of TU cannot deal with the scarcity of labeled tabular data. In addition, these methods primarily focus on the textual content within the table, disregarding the inherent topological information of the table. This can lead to a misunderstanding of the tabular semantics. In this paper, we propose TabPrompt, a new framework to tackle the above challenges. Prompt-based learning has gained popularity due to its exceptional performance in few-shot learning. Thus, we introduce prompt-based learning to handle few-shot TU. Furthermore, Graph Contrastive Learning (Graph CL) demonstrates remarkable capabilities in capturing topological information, making Graph Neural Networks an ideal method for encoding tables. Hence, we develop a novel Graph CL method tailored to tabular data. This method serves as the pretext task during the pre-training phase, allowing the generation of vector representations that incorporate the table's topological information. The experimental results of outperforming all strong baselines demonstrate the strength of our method in fewshot table understanding tasks.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
4 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding cites TURL: Table Understanding through Representation Learning
- HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding cites TUTA: Tree-based Transformers for Generally Structured Table Pre-training
- HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding cites TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data
- TableFormer: Robust Transformer Modeling for Table-Text Encoding cites TURL: Table Understanding through Representation Learning
- TableFormer: Robust Transformer Modeling for Table-Text Encoding cites TUTA: Tree-based Transformers for Generally Structured Table Pre-training
- HInT: Hypergraph Infusion at the Structural Layers Improves Table Understanding cites TableFormer: Robust Transformer Modeling for Table-Text Encoding
- HInT: Hypergraph Infusion at the Structural Layers Improves Table Understanding cites TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data
- TUTA: Tree-based Transformers for Generally Structured Table Pre-training cites TURL: Table Understanding through Representation Learning
- PixT3: Pixel-based Table-To-Text Generation cites TUTA: Tree-based Transformers for Generally Structured Table Pre-training
Cited by top-tier papers4
- PixT3: Pixel-based Table-To-Text GenerationIñigo Alonso, Eneko Agirre, Mirella LapataACL 2024 · 2 citations
- HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table UnderstandingRihui Jin, Yu Li, Guilin Qi, Nan Hu et al.AAAI 2025 · 1 citation
- FollowTable: A Benchmark for Instruction-Following Table RetrievalRihui Jin, Yuchen Lu, Ting Zhang, Jun Wang et al.SIGIR 2026
- HInT: Hypergraph Infusion at the Structural Layers Improves Table UnderstandingWonjin Lee, Soomi Jeong, Kwang In KimICML 2026
Builds on9
- TURL: Table Understanding through Representation LearningXiang Deng, Huan Sun, Alyssa Lees, You Wu et al.VLDB 2021 · 2,406 citations
- TableFormer: Robust Transformer Modeling for Table-Text EncodingJingfeng Yang, Aditya Gupta, Shyam Upadhyay, Luheng He et al.ACL 2022 · 145 citations
- TUTA: Tree-based Transformers for Generally Structured Table Pre-trainingZhiruo Wang, Haoyu Dong, Ran Jia, Jia Li et al.KDD 2021 · 88 citations
- How Should Pre-Trained Language Models Be Fine-Tuned Towards Adversarial Robustness?Xinshuai Dong, Anh Tuan Luu, Min Lin, Shuicheng Yan et al.NeurIPS 2021 · 80 citations
- TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular DataLun Du, Fei Gao, Xu Chen, Ran Jia et al.KDD 2021 · 58 citations
Related papers
- Trompt: Towards a Better Deep Neural Network for Tabular DataKuan-Yu Chen, Ping-Han Chiang, Hsin-Rung Chou, Ting-Wei Chen et al.ICML 2023 · 42 citations
- HGPrompt: Bridging Homogeneous and Heterogeneous Graphs for Few-Shot Prompt LearningXingtong Yu, Yuan Fang, Zemin Liu, Xinming ZhangAAAI 2024 · 68 citations
- TAROT: Task-Adaptive Refinement of LLM-prior Graphs for Few-shot Tabular LearningRuxue Shi, Yili Wang, Mengnan Du, Hangting Ye et al.KDD 2026
- MultiGPrompt for Multi-Task Pre-Training and Prompting on GraphsXingtong Yu, Chang Zhou, Yuan Fang, Xinming ZhangWWW 2024 · 65 citations
- Language Models are Weak LearnersHariharan Manikandan, Yiding Jiang, J. Zico KolterNeurIPS 2023 · 32 citations
