A Modular LLM Framework for Explainable Price Outlier Detection

Shadi Sartipi; John Wu; Sina Ghotbi; Nikhita Vedula; Shervin Malmasi

cs.CLMar 21, 2026

A Modular LLM Framework for Explainable Price Outlier Detection

Authors: Shadi Sartipi, John Wu, Sina Ghotbi, Nikhita Vedula, Shervin Malmasi

Organizations: Amazon.com, Inc.

Abstract

Detecting product price outliers is important for retail and e-commerce stores as erroneous or unexpectedly high prices adversely affect competitiveness, revenue, and consumer trust. Classical techniques offer simple thresholds while ignoring the rich semantic relationships among product attributes. We propose an agentic Large Language Model (LLM) framework that treats outlier price flagging as a reasoning task grounded in related product detection and comparison. The system processes the prices of target products in three stages: (i) relevance classification selects price-relevant similar products using product descriptions and attributes; (ii) relative utility assessment evaluates the target product against each similar product along price influencing dimensions (e.g., brand, size, features); (iii) reasoning-based decision aggregates these justifications into an explainable price outlier judgment. The framework attains over 75% agreement with human auditors on a test dataset, and outperforms zero-shot and retrieval based LLM techniques. Ablation studies show the sensitivity of the method to key hyper-parameters and testify on its flexibility to be applied to cases with different accuracy requirement and auditor agreements.

Figures & tables

Figure 1: Overview of our proposed agentic framework. We investigate four attribute selection configurations with the Utility Agent (Generic, Static, Dynamic, or W-Dynamic), detailed further in Section 3.3 .

Panel of Figure 2 — Figure 2: (a) The target 4-quadrant framework showing ( left ) the informative zones and ( right ) relevant products shown on uninformative zones along with utility trade-off padding. (b) The sample of the decision showing the ( left ) correct price, and ( right ) outlier price.

Baseline Comparison
Approach	SS F1 $\uparrow$	SS Pre. $\uparrow$	SS Rec. $\uparrow$	Agreement $\uparrow$	EC F1 $\uparrow$	EC Pre. $\uparrow$	EC Rec. $\uparrow$	Outlier Rate
Zero Shot	0.43	0.50	0.38	68%	0.50	0.33	1.0	15.1%
RAG	0.42	0.31	0.63	23.0%	0.29	0.17	1.0	40.9%
Agentic	0.55	1.00	0.39	76.3%	0.50	0.33	1.0	7.8%
Agentic - Price Paddings (@7 Neighbors)
30%	0.71	0.67	0.75	52.6%	0.29	0.17	1.0	14.9%

Table 1: Comparing our agentic approach with baselines and hyper-parameter variations on the silver set (SS), one-sided annotated set (measured by agreement score), edge cases (EC), and outlier rate on the unannotated set. ‘Pre.’ and ‘Rec.’ stand for Precision and Recall respectively.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Configuration	SS F1 $\uparrow$	SS Pre. $\uparrow$	SS Rec. $\uparrow$	Agreement $\uparrow$	EC F1 $\uparrow$	EC Pre. $\uparrow$	EC Rec. $\uparrow$	Outlier Rate
Amazon Nova Pro	0.46	0.3	1.0	3.0%	0.29	0.16	0.28	26.2%
Claude 3.5 Sonnet	0.67	0.54	0.88	42.1%	0.50	0.33	1.0	10.4%
Claude 4 Sonnet	0.54	0.39	0.88	21.6%	0.33	0.20	1.0	24.2%

Appendix

Table 2: Performance comparison across different LLM models. Padding and Neighbor are set to LLM-Based and 7, respectively.

Figure 3: Example for the Utility Agent analysis.

# of Products	Time (Claude 3.5)	Time (Human)	Cost (Claude)	Cost (Human)
10	0.27	3.33	$1.05	$33.33
1,000	27.03	333.33	$105.30	$3,333.33
400,000,000	10,810,810.81	133,333,333.33	$42,120,000.00	$1,333,333,333.33

Appendix

Table 3: Comparison of Analysis Costs: Claude 3.5 vs. Human. Time is measured in hours.

Anomaly Detection

Anomaly detection finds rare or unexpected patterns in data, from defective products and system faults to fraud and security threats.

See all 652 papers

LLP: LLM-Based Product Pricing in E-commerce
Oct 10, 2025Hairu Wang, Sheng You, Qiheng Zhang +5E-Commerce LLM Post-Training
Xianyu of Alibaba, Hangzhou, China · University of Science and Technology of China, Suzhou, China
Algorithmic Collusion by Large Language Models
Mar 31, 2024Sara Fish, Yannai A. Gonczarowski, Ran I. ShorrerPrompt Sensitivity Algorithmic Collusion
†School of Engineering and Applied Sciences, Harvard University · ‡Department of Economics and Department of Computer Science, Harvard University · §Department of Economics, Penn State University
Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer
Sep 23, 2026Davood Wadi, Yu MaAgentic Commerce LLM Decision-Making
Desautels Faculty of Management, McGill University, Montreal, Quebec, Canada

Oct 10, 2025J/K move · Enter open · S save

LLP: LLM-Based Product Pricing in E-commerce

Hairu Wang, Sheng You, Qiheng Zhang, Xike Xie+4

Xianyu of Alibaba, Hangzhou, China · University of Science and Technology of China, Suzhou, China

E-Commerce LLM Post-Training Dynamic Pricing

Open paper AskPDF arXiv

Unlike Business-to-Consumer e-commerce platforms (e.g., Amazon), inexperienced individual sellers on Consumer-to-Consumer platforms (e.g., eBay) often face significant challenges in setting prices for their second-hand products efficiently. Therefore, numerous studies have been proposed for automating price prediction. However, most of them are based on static regression models, which suffer from poor generalization performance and fail to capture market dynamics (e.g., the price of a used iPhone decreases over time). Inspired by recent breakthroughs in Large Language Models (LLMs), we introduce LLP, the first LLM-based generative framework for second-hand product pricing. LLP first retrieves similar products to better align with the dynamic market change. Afterwards, it leverages the LLMs' nuanced understanding of key pricing information in free-form text to generate accurate price suggestions. To strengthen the LLMs' domain reasoning over retrieved products, we apply a two-stage optimization, supervised fine-tuning (SFT) followed by group relative policy optimization (GRPO), on a dataset built via bidirectional reasoning. Moreover, LLP employs a confidence-based filtering mechanism to reject unreliable price suggestions. Extensive experiments demonstrate that LLP substantially surpasses existing methods while generalizing well to unseen categories. We have successfully deployed LLP on Xianyu\footnote{Xianyu is China's largest second-hand e-commerce platform.}, significantly outperforming the previous pricing method. Under the same 30% product coverage, it raises the static adoption rate (SAR) from 40% to 72%, and maintains a strong SAR of 47% even at 90% recall.

Abstract

Anomaly Detection

Explore similar work

LLP: LLM-Based Product Pricing in E-commerce

Algorithmic Collusion by Large Language Models

Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer