ST-Raptor Intelligent Table question-answering System Based on Tree Framework

The tree-based intelligent table question answering system (ST-Raptor) was jointly proposed by Shanghai Jiao Tong University and other universities in August 2025. The relevant research results were published in the paper "ST-Raptor: LLM-Powered Semi-Structured Table Question Answering".

ST-Raptor is a tree-based framework for answering semi-structured table questions using large language models. Researchers introduced a hierarchical orthogonal tree (HO-Tree) to capture the complex layout of semi-structured tables. They then defined a set of basic tree operations to guide LLMs in performing common question-answering tasks. Given a user question, ST-Raptor decomposes it into simpler sub-questions, generates corresponding tree-based operation flows, and performs operation-table alignment to accurately execute the flow.