Automated Domain Model Extraction from Semi-Structured Documents

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Building detailed domain models from scratch is time-consuming and cumbersome, especially in complex domains where no single person is an authoritative expert, and existing approaches fail to leverage fragmented domain models in documents created by word processors.

Innovation Solution

The system automatically harvests documents to separate content from presentation, identifies candidate model elements within and across document types, consolidates these to learn a global model, and allows for manual review to resolve domain-specific ambiguities, using indicative words for classification and concept extraction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If domain models are built from scratch manually, then model quality and accuracy can be ensured, but the time and effort required becomes excessively large

Engineering Contradiction:

Improvemodel qualityVSAvoidtime to build model

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by automatically harvesting and analyzing existing documentation before model construction. It pre-processes documents to extract domain concepts, relationships, and terminology, creating a foundation that significantly reduces the time required for subsequent manual model building while maintaining quality through structured extraction methods

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary automated system that acts as a bridge between existing documentation and the final domain model. This intermediary tooling harvests information from documents, extracts domain concepts and relationships, and presents structured data to modelers, reducing both time and manual effort while preserving model quality

Inventive Principle:

Principle #24Intermediary (Mediator)

2Manufacturing precision

If domain experts are consulted to build models, then domain accuracy is improved, but the complexity of coordinating multiple experts and locations increases

Engineering Contradiction:

Improvedomain accuracyVSAvoidcoordination complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system segments the model building process into independent, automated stages: document harvesting, concept extraction, relationship identification, and model generation. This segmentation allows different teams to work on separate document sets independently, with the system automatically integrating results, thereby reducing coordination complexity while maintaining domain accuracy through systematic processing

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent creates copies of domain knowledge from existing documentation across multiple locations and teams. By harvesting and extracting information from various document sources into a unified model structure, it replicates domain expertise without requiring constant expert coordination, reducing complexity while preserving accuracy

Inventive Principle:

Principle #26Copying

3Productivity

If automated document processing is used, then productivity is improved, but the precision of domain concept extraction may deteriorate

Engineering Contradiction:

Improvemodel building speedVSAvoidconcept extraction accuracy

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The system incorporates feedback mechanisms where extracted domain concepts and relationships are validated against the source documentation and refined iteratively. Automated processing identifies candidate concepts and relationships, which are then reviewed and refined based on feedback from the documentation context, maintaining high precision while achieving productivity through automation

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent replaces manual mechanical analysis of documents with automated text processing and information extraction systems. These systems use computational methods to identify domain concepts, relationships, and terminology with high precision, achieving both productivity improvement through automation and accuracy maintenance through sophisticated NLP techniques

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS10089390B2System and method to extract models from semi-structured documents

Publication Date: 2018.10.02 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Systems and associated methods for automated and semi-automated building of domain models for documents are described. Embodiments provide an approach to discover an information model by mining documentation about a particular domain captured in the documents. Embodiments classify the documents into one or more types corresponding to concepts using indicative words, identify candidate model elements (concepts) for document types, identify relationships both within and across document types, and consolidate and learn a global model for the domain.