Transaction String Classification Using Category Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Transaction strings in financial transactions exhibit significant variability, making it difficult to categorize and process them effectively for further analysis and verification tasks such as income verification and benefit administration.
Innovation Solution
A machine learning system is employed to classify transaction strings into categories like deposit sources using algorithms like decision trees, boosting, and neural networks, and preprocesses the data to reduce variability, enabling accurate classification and identification of deposit sources and counterparties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional transaction string processing is used, then financial institutions can maintain their own formatting and content standards, but transaction categorization becomes difficult due to significant variability
Solution Approach 1:
The patent transforms transaction strings by changing their parameters through normalization processes. It extracts key features from variable transaction strings and converts them into standardized numerical representations, allowing diverse transaction formats to be processed uniformly while maintaining their essential characteristics
Solution Approach 2:
The patent introduces machine learning models as intermediary components between raw transaction strings and categorization outcomes. These models act as mediators that learn to map variable transaction formats to consistent categories, bridging the gap between institutional formatting flexibility and uniform processing requirements
2Reliability
If machine learning models are trained on raw transaction strings, then the models can learn from real data, but the training data contains significant noise and variability that reduces model performance
Solution Approach 1:
The patent applies preprocessing actions before the main training process. It performs feature extraction, normalization, and cleaning of transaction strings prior to feeding them into machine learning models, removing noise and variability that would otherwise degrade model performance and reduce data quality
Solution Approach 2:
The patent extracts essential features from noisy raw transaction strings while discarding irrelevant variability. It separates signal from noise by identifying and extracting key transaction characteristics that are relevant for categorization and income verification, removing institutional formatting variations
Data Source
AI summary
Provided are systems and methods which can use machine learning to draw additional inferences about transaction records from transaction strings. The inferred data can be used to build a classification model configured to map transaction string to predefined categories. In one example, a method may include receiving a file comprising transaction strings corresponding to a plurality of transaction records, executing the machine learning model on the transaction strings to identify a plurality of categories associated with the transaction strings, generating a classifier model that comprises patterns of keywords from the transactions strings mapped to the plurality of identified categories, respectively, and storing the classifier model in the data store.