An Comparative Analysis about KYC on a Recommendation System Toward Agentic Recommendation System
Junjie H. Xu
TL;DR
This work investigates an agentic recommender system that treats Know Your Customer (KYC) data as a dynamic memory/context module to drive cross-domain personalization and exploration. By combining multimodal content representations, social-graph signals, and a Global Logic layer, the approach progressively deepens KYC from cold-start to rich personal context and social circles, achieving higher $nDCG$ and serendipity across Ad, News, Gossip, Sharing, and Tech. Experimental results show substantial gains, with the deepest KYC plus Circles configuration delivering the strongest improvements, especially in Sharing and Tech domains. The findings suggest that, beyond accuracy, agentic, context-aware, cross-domainRecommendation with exploration mechanisms can break information silos and unlock long-tail content discovery, offering a practical path toward next-generation recommender systems.
Abstract
This research presents a cutting-edge recommendation system utilizing agentic AI for KYC (Know Your Customer in the financial domain), and its evaluation across five distinct content verticals: Advertising (Ad), News, Gossip, Sharing (User-Generated Content), and Technology (Tech). The study compares the performance of four experimental groups, grouping by the intense usage of KYC, benchmarking them against the Normalized Discounted Cumulative Gain (nDCG) metric at truncation levels of $k=1$, $k=3$, and $k=5$. By synthesizing experimental data with theoretical frameworks and industry benchmarks from platforms such as Baidu and Xiaohongshu, this research provides insight by showing experimental results for engineering a large-scale agentic recommendation system.