Multimodal Banking Dataset: Understanding Client Needs through Event Sequences

2024-09-26

arXiv.org

AI Summary

The financial domain has long lacked large-scale, multi-source, real-world event sequence datasets, hindering research on multimodal temporal modeling. To address this, we introduce MBD—the first industrial-grade, open-source multimodal banking event sequence dataset—comprising 1.5 million corporate clients, 950 million transactions, 1 billion geolocation records, 5 million customer-service dialogue embeddings, and aggregated product-purchase data. We propose three core techniques: cross-modal event alignment, differential-privacy-based anonymization, and multimodal temporal encoding. Further, we define two benchmark tasks—marketing response prediction and client matching. Extensive experiments demonstrate that our multimodal baselines significantly outperform unimodal counterparts, effectively overcoming modeling bottlenecks imposed by financial data silos and privacy constraints. MBD establishes a new infrastructure for reproducible, scalable research in financial intelligent decision-making.

Technology Category

• Machine Learning: Multimodal Learning
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data
• Computer Vision: Multi-modal Vision

Application Category

• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Financial organizations collect a huge amount of data about clients that typically has a temporal (sequential) structure and is collected from various sources (modalities). Due to privacy issues, there are no large-scale open-source multimodal datasets of event sequences, which significantly limits the research in this area. In this paper, we present the industrial-scale publicly available multimodal banking dataset, MBD, that contains more than 1.5M corporate clients with several modalities: 950M bank transactions, 1B geo position events, 5M embeddings of dialogues with technical support and monthly aggregated purchases of four bank's products. All entries are properly anonymized from real proprietary bank data. Using this dataset, we introduce a novel benchmark with two business tasks: campaigning (purchase prediction in the next month) and matching of clients. We provide numerical results that demonstrate the superiority of our multi-modal baselines over single-modal techniques for each task. As a result, the proposed dataset can open new perspectives and facilitate the future development of practically important large-scale multimodal algorithms for event sequences. HuggingFace Link: https://huggingface.co/datasets/ai-lab/MBD Github Link: https://github.com/Dzhambo/MBD

Problem

• Lack of large open-source multimodal banking datasets for deep learning
• Need for anonymized real-world client data from multiple sources
• Absence of benchmarks for financial multimodal sequence analysis tasks

Innovation

• First industrial-scale multimodal banking dataset
• Anonymized data preserving significant information
• Fusion baselines outperform single-modal techniques