AI Polymer Structural Formula Recognition for Copolymer Brackets

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems struggle to accurately recognize polymer molecular structural formulas, particularly those of copolymers with multiple monomers, due to their representation in image form, making it difficult to identify and select necessary information.

Innovation Solution

A system utilizing artificial intelligence with a first model to detect brackets and subscripts, and a second model to output cluster data, including group information, to accurately recognize polymer molecular structures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Shape

If molecular structural formulas are provided in image form, then visual representation is improved, but identification and selection difficulty increases

Engineering Contradiction:
Improvevisual representationVSAvoididentification difficulty
Core Design Contradiction:
ShapeVSDifficulty of detecting and measuring

Solution Approach 1:

The patent replaces manual visual inspection and identification of molecular structures with an automated AI-based image processing system. The system uses neural networks to detect, recognize, and convert molecular structural formula images into searchable data formats, eliminating the need for manual identification while preserving the visual representation benefits.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Loss of information

If polymer molecular structures are represented with brackets and subscripts, then copolymer information is improved, but recognition accuracy deteriorates

Engineering Contradiction:
Improvecopolymer informationVSAvoidrecognition accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent segments the complex polymer molecular structure into distinct components: monomer units, brackets indicating repetition, and subscripts showing quantities. The AI system processes each element separately, detecting brackets and subscripts as specific features, then reconstructs the complete polymer structure with accurate copolymer information preserved.

Inventive Principle:
Principle #1Segmentation

3Ease of operation

If general search methods are used, then search simplicity is improved, but search effectiveness deteriorates

Engineering Contradiction:
Improvesearch simplicityVSAvoidsearch effectiveness
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent introduces an intermediary AI processing system that converts molecular structure images into standardized data formats (such as SMILES notation). This intermediary step enables the use of simple text-based search methods while maintaining high search effectiveness, as the converted data can be efficiently queried and compared.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4618047A1System, method, and program for recognizing polymer molecular structural formula by using artificial intelligence
Publication Date: 2025.09.17 LG MANAGEMENT DEV INST CO LTD
  • EP4618047A1 patent drawingFigure 1
  • EP4618047A1 patent drawingFigure 2
  • EP4618047A1 patent drawingFigure 3

AI summary

A system, method, and program for recognizing a polymer molecular structural formula that use artificial intelligence are disclosed. The system for recognizing the polymer molecular structural formula using artificial intelligence includes at least one processor, and at least one memory storing a command or information that cause the at least one processor to perform an operation, wherein the operation performed by the command or the information includes detecting a polymer molecular structural formula image to generate detection data including information about a bracket and a subscript, and inputting the detection data to each of a first model and a second model to output first cluster data from the first model and to output second cluster data including group information about the bracket and the subscript and including information different from the first cluster data from the second model.