analiza aiqbench
HANIA introduces a planner-guided framework that uses multimodal graphs to filter noisy evidence for grounded question answering, eliminating the need for dataset-specific fine-tuning. By coordinating descriptive and relational data through finite-state planning, the system improves accuracy while maintaining efficiency on benchmarks like ScienceQA. This approach directly addresses core challenges in multimodal reasoning and evidence selection within large language models.
HANIA: Planner-Guided Multimodal Graph Evidence Selection for Grounded Question Answering
arXiv:2608.29088v1 Announce Type: new Abstract: Multimodal question answering remains sensitive to noisy, incomplete, and weakly grounded evidence. Long unstructured contexts can introduce redundancy and encourage unsupported generation, while flat retrieval may overlook relations needed for multi-step reasoning. We present HANIA, a planner-guided multimodal graph framework for evidence-grounded question answering. HANIA processes the supplied image and text using a frozen vision-language model to extract concise question-relevant visual evidence with explicit abstention. It then constructs an input-grounded multimodal graph and applies a two-group finite-state planner to coordinate descriptive and relational evidence. Coverage-aware pruning retains a compact evidence set based on relevance, graph confidence, concept coverage, and modality diversity. The selected passages, visual statements, and graph triples are provided to a frozen instruction-tuned decoder. We evaluate HANIA on ScienceQA using answer accuracy, evidence-filtering quality, evidence-budget sensitivity, and efficiency. The results show that structured evidence planning and compact graph-guided retrieval can support competitive multimodal question answering without target-dataset fine-tuning or iterative retrieval. The code is available at https://github.com/Zafar-southeast/HANIA.
Abstract:Multimodal question answering remains sensitive to noisy, incomplete, and weakly grounded evidence. Long unstructured contexts can introduce redundancy and encourage unsupported generation, while flat retrieval may overlook relations needed for multi-step reasoning. We present HANIA, a planner-guided multimodal graph framework for evidence-grounded question answering. HANIA processes the supplied image and text using a frozen vision-language model to extract concise question-relevant visual evidence with explicit abstention. It then constructs an input-grounded multimodal graph and applies a two-group finite-state planner to coordinate descriptive and relational evidence. Coverage-aware pruning retains a compact evidence set based on relevance, graph confidence, concept coverage, and modality diversity. The selected passages, visual statements, and graph triples are provided to a frozen instruction-tuned decoder. We evaluate HANIA on ScienceQA using answer accuracy, evidence-filtering quality, evidence-budget sensitivity, and efficiency. The results show that structured evidence planning and compact graph-guided retrieval can support competitive multimodal question answering without target-dataset fine-tuning or iterative retrieval. The code is available at this https URL.
10 pages, 1 figure. Accepted at Graph-enhanced LLMs for trustwOrthy Web data management (GLOW), ISWC 2026 Workshops, Bari, Italy
Artificial Intelligence (cs.AI)
arXiv:2608.29088 [cs.AI]
(or arXiv:2608.29088v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2608.29088 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 274 cuvinte
Subiecte conexe
VISPATH: Visual-Intent-Guided Path Reasoning for Multimodal Knowledge Graph Question Answering
arXiv · 21 sept. 2026
[AINews] not much happened today
Latent Space · 10 sept. 2026
K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs
arXiv · 23 iul. 2026
SAFE-G: Structure-aware Faithful Evidence-guided Generation for Knowledge-based Visual Question Answering
arXiv · 25 aug. 2026