Summary of Hegta: Leveraging Heterogeneous Graph-enhanced Large Language Models For Few-shot Complex Table Understanding, by Rihui Jin et al.
HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding
by Rihui Jin, Yu Li, Guilin Qi, Nan Hu, Yuan-Fang Li, Jiaoyan Chen, Jianan Wang, Yongrui Chen, Dehai Min, Sheng Bi
First submitted to arxiv on: 28 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Databases (cs.DB); Multimedia (cs.MM)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a framework called HGT to tackle table understanding (TU) with limited training data. The main challenge is the scarcity of manually labeled tables and the complexity of table structures. To address this, HGT combines a heterogeneous graph (HG) with a large language model (LLM). The LLM is trained using soft prompts and instruction turning, allowing it to align with table semantics. Additionally, HGT employs a multi-task pre-training scheme that involves three novel self-supervised HG objectives. Experimental results demonstrate the effectiveness of HGT in outperforming state-of-the-art models for few-shot complex TU on several benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps computers better understand tables with limited training data. One problem is that there aren’t many labeled tables, and tables can be very complicated. The researchers created a new system called HGT to solve this issue. It combines two ideas: a special type of graph (HG) and a large language model (LLM). The LLM learns to understand table meanings by aligning with the HG’s knowledge. The system also uses a new way to train models, which involves multiple tasks. This approach was tested and showed that it can do better than other methods for understanding complex tables. |
Keywords
» Artificial intelligence » Few shot » Large language model » Multi task » Self supervised » Semantics
