Automated SDG Bond Evaluation Using ML Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current evaluation methods for SDGs bonds are costly and labor-intensive, leading to a limited market and potential misclassification of bonds, as they rely on manual certification and relative evaluation, lacking flexibility and detail.
Innovation Solution
An information processing apparatus and method that performs evaluation by comparing a text describing an evaluation target to a comparison target satisfying predetermined criteria, using supervised learning classifiers to assess similarity across multiple sub-goals and ideal patterns, enabling both relative and absolute evaluations, and reducing manual work through automated screening and rating processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual certification and relative evaluation methods are used, then evaluation quality can be maintained through expert judgment, but evaluation costs increase and productivity decreases
Solution Approach 1:
The patent introduces an automated evaluation system that acts as an intermediary between manual expert evaluation and bond certification. The system uses machine learning models trained on historical expert evaluation data to perform automated screening and rating, replicating expert judgment while significantly reducing costs and time requirements.
Solution Approach 2:
The patent creates a digital copy of the expert evaluation process by training machine learning models on historical expert judgment data. The models learn to replicate expert evaluation patterns and can automatically assess bonds with reliability comparable to manual expert review, thereby scaling productivity without sacrificing quality.
2Measurement precision
If manual evaluation processes are used, then detailed assessment can be performed, but the complexity of the evaluation system increases and automation decreases
Solution Approach 1:
The patent divides the evaluation system into distinct modular components: data collection module, feature extraction module, machine learning model training module, automated screening module, and rating module. Each module performs a specific function, making the overall complex system manageable and maintainable while enabling detailed evaluation through coordinated operation of specialized components.
Solution Approach 2:
The patent replaces manual mechanical evaluation processes with automated computational systems. Machine learning algorithms automatically extract features from bond documentation and perform assessments, substituting human expert manual analysis with algorithmic processing that maintains detail while reducing operational complexity.
3Ease of manufacture
If relative evaluation based on existing services is used, then evaluation can be performed using available data, but adaptability to new evaluation criteria decreases
Solution Approach 1:
The patent implements dynamic evaluation criteria that can be adjusted and updated without requiring complete system redesign. The machine learning models are trained on configurable criteria sets, allowing the evaluation system to adapt to new standards, regulations, or investment preferences by simply updating the training data and retraining models, thereby maintaining feasibility while enhancing flexibility.
Solution Approach 2:
The patent enables flexibility by allowing evaluation parameters and criteria to be changed independently. The system can modify weighting factors, threshold values, and criterion definitions to accommodate different evaluation scenarios while maintaining the underlying automated evaluation infrastructure, thus preserving ease of implementation while gaining adaptability.
4Productivity
If automated evaluation using machine learning is implemented, then productivity increases and costs decrease, but measurement precision may be compromised compared to expert judgment
Solution Approach 1:
The patent performs preliminary training of machine learning models using extensive historical evaluation data and expert judgments before deployment. This preliminary action allows the models to learn from numerous examples and achieve high accuracy. The models are validated against held-out test sets to ensure they meet precision requirements before being used for automated evaluation, thereby maintaining accuracy while enabling high productivity.
Data Source
AI summary
An information processing apparatus includes an evaluation section that inputs to a first classifier generated by supervised learning that uses a feature amount extracted from text in which a characteristic of a comparison target evaluated to satisfy a predetermined evaluation criteria defined by a combination of multiple sub goals of the predetermined goal is described, a feature amount extracted from the text in which the characteristic of the evaluation target is described, to acquire a similarity between the evaluation target and the comparison target for each of the sub goals. Further, the evaluation section evaluates, in reference to the similarity between the evaluation target and the comparison target for each of the sub goals and an ideal pattern for each of the predetermined evaluation criteria defined by a combination of the multiple sub goals, the similarity between the ideal pattern for each of the predetermined evaluation criteria and evaluation target.