Financial Transaction Retrieval and Contextual Evidence for Knowledge-Grounded Reasoning

2026-03-16

AI Summary

This work addresses the challenge that general-purpose large language models struggle to effectively model highly sequential and structured financial transaction data, while existing specialized models rely heavily on extensive labeled data and exhibit limited generalization. To overcome these limitations, the authors propose FinTRACE, a novel architecture that introduces a retrieval-first paradigm: it transforms raw transaction records into reusable feature representations, constructs a behavior knowledge base via rule-based detectors, and leverages a retrieval mechanism to supply contextual evidence for downstream tasks. Notably, FinTRACE is the first to incorporate retrieved behavioral patterns into instruction tuning of large language models, aligning structured memory with task objectives. Evaluated on multiple public and industrial benchmarks, FinTRACE achieves state-of-the-art performance under low-supervision settings, improving zero-shot churn prediction MCC from 0.19 to 0.38 and reaching 0.40 MCC with only 16 labeled samples.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Nowadays, success of financial organizations heavily depends on their ability to process digital traces generated by their clients, e.g., transaction histories, gathered from various sources to improve user modeling pipelines. As general-purpose LLMs struggle with time-distributed tabular data, production stacks still depend on specialized tabular and sequence models with limited transferability and need for labeled data. To address this, we introduce FinTRACE, a retrieval-first architecture that converts raw transactions into reusable feature representations, applies rule-based detectors, and stores the resulting signals in a behavioral knowledge base with graded associations to the objectives of downstream tasks. Across public and industrial benchmarks, FinTRACE substantially improves low-supervision transaction analytics, doubling zero-shot MCC on churn prediction performance from 0.19 to 0.38 and improving 16-shot MCC from 0.25 to 0.40. We further use FinTRACE to ground LLMs via instruction tuning on retrieved behavioral patterns, achieving state-of-the-art LLM results on transaction analytics problems.

Problem

• financial transaction retrieval
• knowledge-grounded reasoning
• low-supervision learning
• tabular data
• behavioral knowledge base

Innovation

• retrieval-first architecture
• behavioral knowledge base
• low-supervision transaction analytics
• instruction tuning
• feature representation