![[2602.12315] AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12315] AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping
Summary
The paper presents AgenticShop, a benchmark for evaluating agentic systems in personalized web shopping, addressing gaps in current evaluation methods.
Why It Matters
As e-commerce grows, effective product curation becomes crucial for enhancing user experience. This research highlights the inadequacies of existing benchmarks and proposes a new framework that emphasizes personalization and realistic shopping scenarios, which is essential for improving online shopping efficiency.
Key Takeaways
- AgenticShop is the first benchmark for personalized product curation in open-web environments.
- Current agentic systems are insufficient in adapting to diverse user preferences.
- The framework includes realistic shopping scenarios and a checklist-driven evaluation for personalization.
Computer Science > Information Retrieval arXiv:2602.12315 (cs) [Submitted on 12 Feb 2026] Title:AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping Authors:Sunghwan Kim, Ryang Heo, Yongsik Seo, Jinyoung Yeo, Dongha Lee View a PDF of the paper titled AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping, by Sunghwan Kim and 4 other authors View PDF HTML (experimental) Abstract:The proliferation of e-commerce has made web shopping platforms key gateways for customers navigating the vast digital marketplace. Yet this rapid expansion has led to a noisy and fragmented information environment, increasing cognitive burden as shoppers explore and purchase products online. With promising potential to alleviate this challenge, agentic systems have garnered growing attention for automating user-side tasks in web shopping. Despite significant advancements, existing benchmarks fail to comprehensively evaluate how well agentic systems can curate products in open-web settings. Specifically, they have limited coverage of shopping scenarios, focusing only on simplified single-platform lookups rather than exploratory search. Moreover, they overlook personalization in evaluation, leaving unclear whether agents can adapt to diverse user preferences in realistic shopping contexts. To address this gap, we present AgenticShop, the first benchmark for evaluating agentic systems on personalized product curation in open-web environment. Crucia...
