Corporate Hierarchy Tagging With ANN Name Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In supply chain management, identifying the ultimate owner of goods shipped by subsidiaries with varying names from their corporate parents is challenging, especially in international transit, as shipping labels often do not clearly indicate the corporate hierarchy.
Innovation Solution
A neural network converts company names into numeric vectors, indexed by euclidean distances, and uses an approximate nearest neighbors (ANN) index to identify the top N nearest neighbors, with a machine learning voting model determining the parent company based on these neighbors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional name matching methods are used to identify parent companies, then the system is simple and easy to operate, but it fails to accurately identify parent companies when subsidiary names differ considerably from the parent name
Solution Approach 1:
The patent replaces traditional mechanical/string-based name matching mechanisms with a neural network-based semantic understanding system. The neural network converts company names into numeric vectors that capture semantic meaning, enabling accurate identification even when names differ significantly, thus resolving the contradiction between accuracy and complexity.
Solution Approach 2:
The patent changes the representation parameter of company names from literal string sequences to numeric vector representations. This parameter transformation allows the system to capture semantic relationships and identify parent companies based on meaning rather than exact name matching, improving accuracy while managing complexity through the ANN index structure.
2Reliability
If exact name matching is used for shipping labels, then the process is fast and simple, but it cannot handle name variations and misspellings in international transit
Solution Approach 1:
The patent performs preliminary action by pre-converting company names into numeric vectors and pre-building the approximate nearest neighbors index during system initialization. This preliminary processing enables fast query responses during actual shipping label matching, improving reliability without significant time loss during operation.
Solution Approach 2:
The patent substitutes exact string matching with neural network-based semantic matching. This substitution allows the system to handle name variations and misspellings reliably while maintaining speed through the optimized ANN search mechanism, resolving the contradiction between reliability and time efficiency.
3Measurement precision
If a comprehensive database of all company names is maintained, then the system can accurately identify parent companies, but the data storage and processing requirements become excessive
Solution Approach 1:
The patent extracts only the essential information needed for identification by converting company names into compact numeric vectors. This extraction process captures the semantic essence of company names while discarding redundant information, significantly reducing the data quantity required while maintaining identification accuracy.
Solution Approach 2:
The patent transforms the data representation from high-dimensional string data to lower-dimensional numeric vectors. This dimensional transformation reduces storage requirements and improves processing efficiency while preserving the semantic information necessary for accurate parent company identification.
Data Source
AI summary
A method for identifying a parent company in a corporate hierarchy is provided. A neural network embeds a number of company names into respective numeric vectors and then indexes the vectors into an approximate nearest neighbors (ANN) index according to euclidean distances between them. In response to a query regarding a parent company of a specified query company, a top N number of nearest neighbor companies to the query company is extracted from the ANN index. A machine learning voting model determines which, if any, of the extracted nearest neighbor companies has a parent company that best corresponds to the query company. If an extracted nearest neighbor company has a parent company that best corresponds to the query company, the parent company is displayed to a user through a user interface.