Item Level Transaction Data Prediction via Merchant Ads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional financial institutions lack access to item-level transaction information, limiting their ability to provide insights and meaningful search capabilities for customers due to the absence of details in transaction data.
Innovation Solution
A system predicts item-level transaction information based on merchant advertisement data by parsing advertisement information using NLP, image recognition, and OCR to generate a price list, matching items with common payment amounts, and updating transaction records with likely item-level data, thereby enhancing data inference and user experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional financial institutions process transaction data, then transaction authentication and processing can be performed, but item-level transaction information remains inaccessible, limiting customer insights and search capabilities
Solution Approach 1:
The patent introduces merchant advertisement information as an intermediary data source to bridge the gap between accessible transaction-level data and inaccessible item-level details. By using advertised product prices as a mediator, the system infers item-level information without direct access to merchant transaction databases, thus resolving the information loss while avoiding the complexity of integrating with merchant systems.
Solution Approach 2:
The patent replaces the mechanical approach of directly accessing merchant transaction systems with an information inference mechanism. Instead of physically or directly electronically accessing item-level data at the source, the system uses computational methods to infer item-level details from aggregated transaction patterns and advertised price lists, reducing system complexity while recovering information.
2Measurement precision
If the system retrieves and parses merchant advertisement information to generate price lists, then item-level data can be inferred, but processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-fetching and storing merchant advertisement information and price lists before they are needed for transaction inference. The system proactively retrieves advertised product data and maintains updated price lists in advance, so when item-level inference is required, the data is already available, significantly reducing processing time while maintaining high accuracy.
3Loss of information
If the system detects transaction patterns and matches items with common payment amounts, then item-level predictions can be made, but the complexity of pattern recognition and matching algorithms increases
Solution Approach 1:
The patent uses copying by creating simplified representations of transaction patterns rather than analyzing raw transaction data directly. The system generates aggregated patterns from multiple transactions that share common characteristics (such as identical payment amounts), effectively copying the essential information structure while filtering out noise. This approach recovers item-level information while keeping the pattern recognition system manageable in complexity.
Data Source
AI summary
Systems as described herein may include predicting item level data based on merchant advertisement information. A transaction pattern may be detected. The merchant advertisement information may be retrieved and parsed to generate a price list. A number of transactions that each shares a common payment amount may be determined and the number may reach a threshold value. Items from the price list may be matched with the common payment amount. The transaction records may be updated to indicate likely item level transaction information. In a variety of embodiments, the likely transaction information may be presented to a user.