Failure is Feedback: History-Aware Backtracking for Agentic Traversal in Multimodal Graphs

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Yun, Joohyung, Lee, Doyup, Han, Wook-Shin
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918321192960000
author Yun, Joohyung
Lee, Doyup
Han, Wook-Shin
author_facet Yun, Joohyung
Lee, Doyup
Han, Wook-Shin
contents Open-domain multimodal document retrieval aims to retrieve specific components (paragraphs, tables, or images) from large and interconnected document corpora. Existing graph-based retrieval approaches typically rely on a uniform similarity metric that overlooks hop-specific semantics, and their rigid pre-defined plans hinder dynamic error correction. These limitations suggest that a retriever should adapt its reasoning to the evolving context and recover intelligently from dead ends. To address these needs, we propose Failure is Feedback (FiF), which casts subgraph retrieval as a sequential decision process and introduces two key innovations. (i) We introduce a history-aware backtracking mechanism; unlike standard backtracking that simply reverts the state, our approach piggybacks on the context of failed traversals, leveraging insights from previous failures. (ii) We implement an economically-rational agentic workflow. Unlike conventional agents with static strategies, our orchestrator employs a cost-aware traversal method to dynamically manage the trade-off between retrieval accuracy and inference costs, escalating to intensive LLM-based reasoning only when the prior failure justifies the additional computational investment. Extensive experiments show that FiF achieves state-of-the-art retrieval on the benchmarks of MultimodalQA, MMCoQA and WebQA.
format Preprint
id arxiv_https___arxiv_org_abs_2602_03432
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Failure is Feedback: History-Aware Backtracking for Agentic Traversal in Multimodal Graphs
Yun, Joohyung
Lee, Doyup
Han, Wook-Shin
Information Retrieval
Open-domain multimodal document retrieval aims to retrieve specific components (paragraphs, tables, or images) from large and interconnected document corpora. Existing graph-based retrieval approaches typically rely on a uniform similarity metric that overlooks hop-specific semantics, and their rigid pre-defined plans hinder dynamic error correction. These limitations suggest that a retriever should adapt its reasoning to the evolving context and recover intelligently from dead ends. To address these needs, we propose Failure is Feedback (FiF), which casts subgraph retrieval as a sequential decision process and introduces two key innovations. (i) We introduce a history-aware backtracking mechanism; unlike standard backtracking that simply reverts the state, our approach piggybacks on the context of failed traversals, leveraging insights from previous failures. (ii) We implement an economically-rational agentic workflow. Unlike conventional agents with static strategies, our orchestrator employs a cost-aware traversal method to dynamically manage the trade-off between retrieval accuracy and inference costs, escalating to intensive LLM-based reasoning only when the prior failure justifies the additional computational investment. Extensive experiments show that FiF achieves state-of-the-art retrieval on the benchmarks of MultimodalQA, MMCoQA and WebQA.
title Failure is Feedback: History-Aware Backtracking for Agentic Traversal in Multimodal Graphs
topic Information Retrieval
url https://arxiv.org/abs/2602.03432