RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA

TL;DR

RELOOP is a structure-aware, iterative retrieval framework for heterogeneous question answering that improves accuracy and efficiency by selectively gathering evidence across text, tables, and knowledge graphs.

Contribution

It introduces a hierarchical sequence representation and a guided, budget-aware iteration process for multi-source, multi-hop retrieval in QA tasks.

Findings

RELOOP achieves higher EM/F1 scores on HotpotQA, HybridQA, and MetaQA datasets.

It unifies processing across text, tables, and KGs without dataset-specific tuning.

RELOOP reduces unnecessary retrieval hops and token usage while maintaining accuracy.

Abstract

Retrieval-augmented generation (RAG) remains brittle on multi-step questions and heterogeneous evidence sources, trading accuracy against latency and token/tool budgets. This paper introduces RELOOP, a structure aware framework using Hierarchical Sequence (HSEQ) that (i) linearize documents, tables, and knowledge graphs into a reversible hierarchical sequence with lightweight structural tags, and (ii) perform structure-aware iteration to collect just-enough evidence before answer synthesis. A Head Agent provides guidance that leads retrieval, while an Iteration Agent selects and expands HSeq via structure-respecting actions (e.g., parent/child hops, table row/column neighbors, KG relations); Finally the head agent composes canonicalized evidence to genearte the final answer, with an optional refinement loop to resolve detected contradictions. Experiments on HotpotQA (text), HybridQA/TAT-QA (table+text), and MetaQA (KG) show consistent EM/F1 gains over strong single-pass, multi-hop, and agentic RAG baselines with high efficiency. Besides, RELOOP exhibits three key advantages: (1) a format-agnostic unification that enables a single policy to operate across text, tables, and KGs without per-dataset specialization; (2) \textbf{guided, budget-aware iteration} that reduces unnecessary hops, tool calls, and tokens while preserving accuracy; and (3) evidence canonicalization for reliable QA, improving answers consistency and auditability.