Carbon-Aware NLP Foundation Model for Scope 3 Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately quantify and manage Scope 3 emissions due to their broad range and complexity, particularly in large organizations with numerous supply-chain partners, and lack scalable solutions for commodity recognition and mapping using natural language processing (NLP) foundation models.
Innovation Solution
A Scope 3 emission system utilizing enterprise financial transaction data, enterprise metadata, and crowd-sourced data to derive embeddings, which are used to build a domain-adapted carbon-aware NLP foundation model with sector-wise spatio-temporal weights, reducing estimation uncertainty through machine learning techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to quantify Scope 3 emissions, then the process can be performed with existing tools, but the accuracy and reliability of emission estimates deteriorate due to the broad range and complexity of Scope 3 sources
Solution Approach 1:
The patent segments Scope 3 emissions into three distinct scopes (Scope 1, Scope 2, and Scope 3) with further categorization into direct, indirect, and other indirect emissions. This segmentation allows the complex emission quantification problem to be broken down into manageable components, each addressed with appropriate measurement and estimation methods, thereby improving accuracy without overwhelming system complexity
Solution Approach 2:
The patent introduces an intermediary classification system that maps various emission sources to standardized categories. This intermediary layer translates diverse raw data from multiple sources into a structured format that can be processed by the emission estimation system, improving measurement precision while managing complexity through standardized interfaces
2Productivity
If manual methods are used for commodity recognition and mapping, then the process can be performed with existing capabilities, but the productivity and scalability deteriorate when dealing with numerous supply-chain partners
Solution Approach 1:
The patent replaces manual mechanical classification methods with automated natural language processing (NLP) systems. The NLP foundation model automatically recognizes and categorizes commodities from unstructured data sources, dramatically increasing processing speed and productivity while maintaining or improving classification accuracy through advanced pattern recognition capabilities
Solution Approach 2:
The patent transforms the commodity classification problem by changing the parameters from manual categorical assignment to automated semantic analysis. The NLP system uses language understanding parameters to dynamically classify commodities based on their textual descriptions, enabling scalable processing of numerous supply-chain partners while maintaining high classification accuracy
3Quantity of substance
If comprehensive data collection is performed across all supply-chain partners, then the quantity of emission data improves, but the loss of time and resources increases due to data aggregation challenges
Solution Approach 1:
The patent implements preliminary action by pre-processing and structuring data at the source before aggregation. Data from supply-chain partners is collected and organized in standardized formats upfront, reducing the time and resources required for later data consolidation and analysis, thereby enabling comprehensive data collection without proportional increases in processing time
Data Source
AI summary
A Scope 3 emission system includes a controller with a microprocessor and a memory. The memory stores executable instructions that when executed by the microprocessor cause the microprocessor to perform a computer-executable method. The method comprises generating embeddings based on enterprise financial transaction data, enterprise metadata, and crowd source data, and generating sector wise carbon-aware spatio-temporal weights indicative of an estimated level of Scope 3 emissions produced by a corresponding commodity sector. The method further includes generating a carbon-aware natural language processing (NLP) foundation model (FM) based on the embeddings and the sector wise carbon-aware spatio-temporal weights. User-generated data indicating at least one target commodity sector and spend data associated with the target commodity sector is input into the NLP FM, and the NLP FM outputs an estimation of the Scope 3 emissions based on the least one target commodity sector and the spend data.