Towards Faithful Industrial RAG: A Reinforced Co-adaptation Framework for Advertising QA

Wenwei Li; Ming Xu; Tianle Xia; Lingxiang Hu; Yiding Sun; Linfang Shang; Liqun Liu; Peng Shu

cs.CLFeb 26, 2026

Towards Faithful Industrial RAG: A Reinforced Co-adaptation Framework for Advertising QA

Authors: Wenwei Li, Ming Xu, Tianle Xia, Lingxiang Hu, Yiding Sun, Linfang Shang, Liqun Liu, Peng Shu,

Organizations: Tencent

Abstract

Industrial advertising question answering (QA) is a high-stakes task in which hallucinated content, particularly fabricated URLs, can lead to financial loss, compliance violations, and legal risk. Although Retrieval-Augmented Generation (RAG) is widely adopted, deploying it in production remains challenging because industrial knowledge is inherently relational, frequently updated, and insufficiently aligned with generation objectives. We propose a reinforced co-adaptation framework that jointly optimizes retrieval and generation through two components: (1) Graph-aware Retrieval (GraphRAG), which models entity-relation structure over a high-citation knowledge subgraph for multi-hop, domain-specific evidence selection; and (2) evidence-constrained reinforcement learning via Group Relative Policy Optimization (GRPO) with multi-dimensional rewards covering faithfulness, style compliance, safety, and URL validity. Experiments on an internal advertising QA dataset show consistent gains across expert-judged dimensions including accuracy, completeness, and safety, while reducing the hallucination rate by 72%. A two-week online A/B test demonstrates a 28.6% increase in like rate, a 46.2% decrease in dislike rate, and a 92.7% reduction in URL hallucination. The system has been running in production for over half a year and has served millions of QA interactions.

Figures & tables

Figure 1: Traditional QA vs. our approach over a shared knowledge base. Given the same user query and knowledge items A, B, C, D, traditional methods often yield incomplete , hallucinated , over-generated , or verbose answers. Our method produces an exact answer that remains complete, faithful, and concise.

Figure 2: System overview. Given a user query $q$ and a private knowledge base $K$ , the retrieval system constructs an evidence set $D$ via two parallel channels: a GraphRAG channel over a high-citation knowledge base $K_{h}$ and a traditional RAG channel with query rewriting and BGE + BM25 hybrid retrieval. Results are merged and deduplicated. The RL-tuned generator then produces a response optimized by GRPO with multi-dimensional rewards for faithfulness, style compliance, safety, and URL validity.

Metric	DeepSeek-V3.2		GPT-5.2		Qwen3-32B		Qwen3-32B-SFT		Qwen3-32B-RL
	Base RAG	GraphRAG	Base RAG	GraphRAG	Base RAG	GraphRAG	Base RAG	GraphRAG	Base RAG	Ours
ROUGE-L $\uparrow$	33.27	37.00	32.82	35.88	29.39	32.96	30.79	33.82	31.40	35.49
Accuracy $\uparrow$	7.82	8.37	7.94	8.39	7.25	7.82	7.50	8.10	7.85	8.26
Completeness $\uparrow$	6.78	7.10	6.70	7.08	6.20	6.66	6.43	6.91	6.46	6.99
Clarity $\uparrow$	8.96	8.99	8.92	8.97	8.52	8.74	8.82	8.94	8.83	8.95
Style $\uparrow$	8.19	8.25	8.14	8.25	7.82	8.03	8.07	8.19	8.27	8.33

Table 1: Main experimental results. Ours refers to Qwen3-32B-RL with GraphRAG. Best results in bold , second best underlined .

Figure 3: Knowledge recall enhancement across Base RAG, GraphRAG, and Parallel retrieval. Effective chunks pre query and recall effectiveness in percent.

Figure 4: Training dynamics of multi-dimensional reward components during RL.

Metric	Base RAG	Ours	$\Delta$
Like Rate (%)	0.21	0.27	+28.6%
Dislike Rate (%)	0.26	0.18	$-$ 30.8%
Avg. Conv. Turns	1.54	1.81	+17.5%

Table 2: Online A/B testing at 50% traffic.

Figure 5: FaithEval generalization: accuracy (%) on Inconsistent, Unanswerable, Counterfactual, and Overall.

Figure 6: Offline evaluation comparison: win/tie/lose distribution across four dimensions.

Metric	Baseline	Ours	$\Delta$
Like Rate (%)	0.21	0.27	+28.6%
Dislike Rate (%)	0.26	0.14	$-$ 46.2%
URL Hallu. (%)	0.0041	0.0003	$-$ 92.7%
Latency (s)	2.5	3.1	+24.0%

Table 3: Online A/B testing results (two weeks, 50% traffic).

Module	Latency (ms)
Query Rewriting	690
GraphRAG Retrieval	852
BGE + BM25 Retrieval	167
Reranking	557
Generation	801
Safety Guardrails	230

Table 4: Latency breakdown by module.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 7: Accuracy on the factual QA evaluation set. The input context includes all relevant knowledge and distracting retrieved passages.

Panel of Figure 8 — Figure 8: Comparison of dialogue responses: the previous online answer (left, red border) contains hallucinated links marked in red, while our answer (right, green border) uses validated links with platform-specific terms replaced by placeholders.

Figure 9: A real dialogue case from our deployed system: the user inquires about restrictions on mini-program landing pages in the financial credit industry; the system responds with a structured, policy-grounded answer covering qualifications, content standards, interaction design, prohibited scenarios, and compliant production suggestions.

Question Answering

Question answering studies how AI systems interpret questions and produce accurate, evidence-grounded answers across text, dialogue, and multimodal data.

See all 260 papers

Adaptive Retrieval, Selective Answers: Evaluating Conformal Factuality in Agentic RAG
Oct 26, 2026Muhammad Aimal Rehman, Chi-Kuang YehRAG Evaluation Agentic RAG
Department of Mathematics & Statistics, Georgia State University, Atlanta, GA, USA
Towards Dependable Retrieval-Augmented Generation Using Factual Confidence Prediction
May 4, 2026Florian Geissler, Francesco Carella, Laura Fieback +1Retrieval-Augmented Generation Factual Consistency Evaluation
Fraunhofer Institute for Cognitive Systems (IKS), Munich, Germany · Volkswagen AG, Wolfsburg, Germany
Faithfulness-QA: A Counterfactual Entity Substitution Dataset for Training Context-Faithful RAG Models
Apr 28, 2026Li Ju, Junzhe Wang, Qi ZhangContextual Faithfulness Retrieval-Augmented Generation
College of Computer Science and Artificial Intelligence, Fudan University

Oct 26, 2026J/K move · Enter open · S save

Adaptive Retrieval, Selective Answers: Evaluating Conformal Factuality in Agentic RAG

Muhammad Aimal Rehman, Chi-Kuang Yeh

Department of Mathematics & Statistics, Georgia State University, Atlanta, GA, USA

RAG Evaluation Agentic RAG Adaptive RAG

Open paper AskPDF arXiv

Retrieval-augmented generation (RAG) can improve factual grounding without guaranteeing that every generated claim is supported. Previous work has studied claim-level conformal factuality in single-stage RAG and, more recently, multi-hop RAG. We examine its behavior in agentic RAG, where a language-model policy adaptively issues follow-up retrieval queries or terminates retrieval before response generation. We retain the retrieval-based relevance-scoring formulation and post-generation conformal filtering and evaluate six model-dataset configurations: GPT-6 Luna and Llama 3.1 8B on HotpotQA, Natural Questions, and TriviaQA. Each configuration uses 60 fixed questions, 100 seeded 30/30 calibration-test partitions, five conformal target levels, and a no-conformal-prediction baseline. Across the six configurations, higher target levels are accompanied by higher observed full-support rates and reduced answer availability. At the 95% target, non-empty response coverage ranges from 17.23% to 60.17%, claim retention from 10.72% to 43.16%, and full support among non-empty outputs from 89.14% to 95.56%. We formulate the adaptive retrieval trajectory explicitly and characterize the reliability-informativeness trade-off when conformal factuality is applied after agent termination.

Abstract

Question Answering

Explore similar work

Adaptive Retrieval, Selective Answers: Evaluating Conformal Factuality in Agentic RAG

Towards Dependable Retrieval-Augmented Generation Using Factual Confidence Prediction

Faithfulness-QA: A Counterfactual Entity Substitution Dataset for Training Context-Faithful RAG Models