Unstructured Document Pattern Recognition for Logistics

Solution Overview

Problem

Current systems in logistics struggle to efficiently process and identify key information from diverse, unstructured documents, leading to difficulties in making informed decisions regarding the purchase, transport, and delivery of goods, and are limited in determining document boundaries within batches.

Innovation Solution

A system comprising a document intake machine and a trade executing machine that scans physical documents, identifies patterns, and generates structured templates for unstructured documents, allowing for the transformation and processing of heterogeneous documents without manual separators, thereby learning the location of key information and applying it to subsequent documents of the same format.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual processing methods are used to handle diverse document formats, then processing accuracy may be maintained, but processing time and resource consumption increase significantly

Engineering Contradiction:

Improvedocument processing speedVSAvoidtime spent on manual document processing

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The system enables unstructured documents to self-configure by automatically generating templates through pattern recognition. The trade executing machine autonomously analyzes document patterns, creates structured templates, and updates the template database without manual intervention, allowing the system to serve itself in template creation and maintenance.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system dynamically adapts document processing parameters by learning from each processed document. It modifies template structures, field positions, and extraction rules based on identified patterns in unstructured documents, enabling the system to adjust to varying document formats and improve processing efficiency over time.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If existing document templates are used for processing, then processing speed is improved, but the system cannot handle new or unstructured document formats

Engineering Contradiction:

Improveability to handle new document formatsVSAvoidcomplexity of template management system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system enables unstructured documents to self-configure by automatically generating templates through pattern recognition. The trade executing machine autonomously analyzes document patterns, creates structured templates, and updates the template database without manual intervention, allowing the system to serve itself in template creation and maintenance.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system performs preliminary pattern analysis on unstructured documents to pre-generate templates before they are needed for bulk processing. By proactively creating templates from sample unstructured documents and storing them in the template database, the system prepares in advance for future processing of similar document types.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If pattern recognition is used to identify document structures, then processing accuracy improves, but computational resources and processing time increase

Engineering Contradiction:

Improveaccuracy of key information identificationVSAvoidcomputational resources for pattern analysis

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system applies partial pattern recognition by focusing analysis on specific critical sections and field names within documents rather than analyzing entire documents uniformly. It identifies and processes only the most important patterns related to key information extraction, reducing overall computational burden while maintaining high accuracy for critical data points.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The system performs preliminary pattern analysis on unstructured documents to pre-generate templates before they are needed for bulk processing. By proactively creating templates from sample unstructured documents and storing them in the template database, the system prepares in advance for future processing of similar document types.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10127209B2Transforming unstructured documents

Publication Date: 2018.11.13 BANK OF AMERICA CORP

AI summary

A system for transforming unstructured documents includes a document intake machine comprising a document scanner. The document intake machine receives a physical trade document from an entity wherein the physical trade document has a format. The document scanner scans the physical trade document to create an electronic file of the physical trade document. A trade-executing machine receives the electronic file and compares the format of the electronic file to a database of known document templates. If the format of the electronic file does not match a known document template, the trade executing machine determines that the electronic file is an unstructured document, identifies a pattern between a field name of the electronic document and data associated with the field name, generates a structured template of the electronic file, by incorporating the identified pattern to transform the unstructured document into a structured document, and updates the database of known document templates with the template of the electronic file.