Predicting Recipient Attributes in B2B Transactions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
B2B electronic transactions lack sufficient information about recipient attributes, hindering providers' efficiency in delivering goods and services, as current techniques fail to reliably determine these attributes.
Innovation Solution
A method and system that utilize historical transaction data to predict recipient attributes by computing provider and recipient features based on known attribute values, employing machine learning models to generate predicted attribute values for recipients with unknown attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If B2B electronic transactions are processed with minimal information, then transaction speed and simplicity are improved, but recipient attribute information becomes insufficient, reducing delivery efficiency
Solution Approach 1:
The system pre-computes provider features from historical transaction data before actual transactions occur. By maintaining a database of provider characteristics and recipient attributes from past transactions, the system can quickly retrieve and match features during new transactions without waiting for information gathering, thus maintaining fast processing speed while having attribute information ready in advance.
Solution Approach 2:
The system creates feature representations (copies) of providers and recipients based on their transaction histories. Instead of storing complete detailed information about each provider and recipient, the system computes condensed feature vectors that capture essential characteristics. These feature copies enable efficient comparison and matching while using minimal storage space and processing resources.
2Ease of operation
If recipient attributes are not declared in transactions, then transaction simplicity is maintained, but provider knowledge of recipient characteristics is limited, reducing delivery efficiency
Solution Approach 1:
The system automatically computes provider and recipient features by analyzing their own historical transaction data without requiring external input or manual declaration. Providers and recipients effectively serve themselves by generating their own feature profiles from their transaction histories, eliminating the need for manual attribute declaration while maintaining information accuracy.
Solution Approach 2:
The system continuously learns from transaction outcomes by incorporating feedback from actual transactions. As more transactions are processed, the feature computations become more accurate as the system refines its understanding of provider and recipient characteristics based on observed behavior patterns, improving reliability over time without requiring explicit attribute declarations.
3Measurement precision
If feature computation is performed for all providers and recipients, then prediction accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The system segments the feature computation process into distinct phases: offline pre-computation of provider features from historical data, and online computation of recipient features during transactions. By dividing the computational task and performing different types of computations at different times, the system achieves high prediction accuracy while managing computational complexity through temporal and functional segmentation.
Data Source
AI summary
The present disclosure provides techniques for recommending vendors using machine learning models. One example method includes receiving electronic transaction data indicative of one or more transactions, identifying, from the one or more transactions, a subset of transactions that are associated with for known attribute values with respect to one or more unique recipients, computing, for each unique provider of the one or more unique providers, a provider feature based on the known attribute values with respect to a subset of the one or more associated unique recipients, computing, for a given recipient indicated in one or more given transactions that are not included in the subset of transactions, a recipient feature based on the provider feature of each unique provider of the one or more associated unique providers, and predicting, based on the recipient feature, a value for the attribute with respect to the given recipient.