Summary of Denoising Table-text Retrieval For Open-domain Question Answering, by Deokhyung Kang et al.
Denoising Table-Text Retrieval for Open-Domain Question Answering
by Deokhyung Kang, Baikjin Jung, Yunsu Kim, Gary Geunbae Lee
First submitted to arxiv on: 26 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 The paper proposes Denoised Table-Text Retriever (DoTTeR) to address challenges in table-text open-domain question answering. The retriever system retrieves relevant evidence from tables and text to answer questions, but previous studies were affected by false-positive labels in training datasets and struggled with reasoning across the table. DoTTeR utilizes a denoised training dataset with fewer false positive labels and integrates table-level ranking information into the retriever to find evidence for complex questions. The approach fine-tunes a rank-aware column encoder to identify minimum and maximum values within a column. Experimental results show that DoTTeR outperforms strong baselines on retrieval recall and downstream QA tasks, demonstrating its effectiveness in handling false positives and reasoning across tables. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves problems with asking questions about tables. It’s hard to find the right answers because training data can be wrong or tricky questions need understanding of patterns in the table. The new approach, called DoTTeR, makes sure the training data is correct and helps find the right evidence for complex questions. It works by making a special kind of computer code that understands how to look at tables and find the important information. This code does better than other approaches on tests, showing it’s effective in solving these problems. |
Keywords
» Artificial intelligence » Encoder » Question answering » Recall
