# From "Who Is This User?" to "What Does This Purchase Mean?": A Deployed Pipeline for Semantic User Profiling at Bank Scale

arXiv: https://arxiv.org/abs/2609.19928 · HTML: https://mlsanity.com/paper/2609.19928/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 37 references that are in the map.
- First posted on arXiv 17 Sep 2026.

## Builds on (most cited first, 22 of 22)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction](https://mlsanity.com/paper/1802.03426/) (2018)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs](https://mlsanity.com/paper/2306.13063/) (2023)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5)](https://mlsanity.com/paper/2203.13366/) (2022)
- [TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation](https://mlsanity.com/paper/2305.00447/) (2023)
- [CoLES: Contrastive Learning for Event Sequences with Self-Supervision](https://mlsanity.com/paper/2002.08232/) (2020)
- [TnT-LLM: Text Mining at Scale with Large Language Models](https://mlsanity.com/paper/2403.12173/) (2024)
- [Transparent and Scrutable Recommendations Using Natural Language User Profiles](https://mlsanity.com/paper/2402.05810/) (2024)
- [Editable User Profiles for Controllable Text Recommendation](https://mlsanity.com/paper/2304.04250/) (2023)
- [Towards a Foundation Purchasing Model: Pretrained Generative Autoregression on Transaction Sequences](https://mlsanity.com/paper/2401.01641/) (2024)
- [You Are What You Bought: Generating Customer Personas for E-commerce Applications](https://mlsanity.com/paper/2504.17304/) (2025)
- [Real-time Attention Based Look-alike Model for Recommender System](https://mlsanity.com/paper/1906.05022/) (2019)
- [Sequences of purchases in credit card data reveal life styles in urban populations](https://mlsanity.com/paper/1703.00409/) (2017)
- [client2vec: Towards Systematic Baselines for Banking Applications](https://mlsanity.com/paper/1802.04198/) (2018)
- [LLM4ES: Learning User Embeddings from Event Sequences via Large Language Models](https://mlsanity.com/paper/2508.05688/) (2025)
- [Fusing Multifaceted Transaction Data for User Modeling and Demographic Prediction](https://mlsanity.com/paper/1712.07230/) (2017)
- [Predicting Multiple Demographic Attributes with Task Specific Embedding Transformation and Attention Network](https://mlsanity.com/paper/1903.10144/) (2019)
- [Improving the Capabilities of Large Language Model Based Marketing Analytics Copilots With Semantic Search And Fine-Tuning](https://mlsanity.com/paper/2404.13077/) (2024)

## Abstract (arXiv)

Per-user LLM inference on transaction histories binds the inference budget linearly to user count, which becomes prohibitive at applied scale. We re-cast attribute inference from per-user to per-transaction-pattern. The pipeline runs in three phases: Resolve abstracts item names with optional web grounding, Profile infers attributes for each frequent pattern, and Tag clusters free-text attributes into a queryable database. In Profile, a single LLM call per pattern emits predefined categorical labels, free-text attributes, and per-attribute prevalence estimates. Because inference runs over patterns rather than users, the budget grows with the pattern count rather than the user count. On the public Open e-commerce corpus, the database is statistically indistinguishable from an LLM that reads each user's raw history directly in AUC across the evaluated attributes, and the prevalence estimates carry discriminative signal between positive and negative users. The pipeline is deployed at a major Japanese bank profiling on the order of tens of millions of users, with close to a three-order-of-magnitude reduction in LLM inference targets versus a per-user pipeline. The code is publicly available on https://github.com/CyberAgentAILab/profiling-agent-open-ecommerce.