Update README.md

README.md

@@ -9,7 +9,11 @@ configs:
   data_files: sqa_query.jsonl
 ---
 
-📄 [Paper](https://huggingface.co/papers/2504.01346) | 👨🏻‍💻 [Code](https://github.com/jiaruzouu/T-RAG)
+📄 [Paper](https://arxiv.org/abs/2504.01346) | 👨🏻‍💻 [Code](https://github.com/jiaruzouu/T-RAG)
+
+## 🔍 Introduction
+
+Retrieval-Augmented Generation (RAG) has become a key paradigm to enhance Large Language Models (LLMs) with external knowledge. While most RAG systems focus on **text corpora**, real-world information is often stored in **tables** across web pages, Wikipedia, and relational databases. Existing methods struggle to retrieve and reason across **multiple heterogeneous tables**.
 
 For MultiTableQA, we release a comprehensive benchmark, including five different datasets covering table fact-checking, single-hop QA, and multi-hop QA: 
 | Dataset              | Link |
@@ -23,32 +27,6 @@ For MultiTableQA, we release a comprehensive benchmark, including five different
 
 MultiTableQA extends the traditional single-table QA setting into a multi-table retrieval and question answering benchmark, enabling more realistic and challenging evaluations.
 
-## Sample Usage
-
-This dataset (`MultiTableQA-SQA`) is part of the larger **MultiTableQA** benchmark. To prepare the full benchmark, you can follow these steps from the official T-RAG GitHub repository.
-
-First, clone the repository and set up the environment:
-
-```bash
-git clone https://github.com/jiaruzouu/T-RAG.git
-cd T-RAG
-
-conda create -n trag python=3.11.9
-conda activate trag
-
-# Install dependencies
-pip install -r requirements.txt
-```
-
-Then, navigate to the `table2graph` directory and run the data preparation script:
-
-```bash
-cd table2graph
-bash scripts/prepare_data.sh
-```
-
-This script will automatically fetch the source tables, apply decomposition (row/column splitting), and generate the benchmark splits, including the `MultiTableQA-SQA` data available in this repository.
-
 ---
 # Citation