Dataset Sufficiency Indicator for AI Model Valuation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Users of AI systems face challenges in determining the sufficiency and value of datasets for modeling purposes, leading to potential overpricing or underpricing of data for transfer, as data holders lack insights into the datasets' usefulness for modeling.

Innovation Solution

A system that generates dataset sufficiency-indicators using a sufficiency database with parameters derived from past dataset transfers, allowing users to estimate the sufficiency and value of datasets for modeling, thereby aiding in informed pricing decisions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If data holders sell datasets without sufficiency evaluation, then data transfer can occur quickly, but the pricing may be inaccurate (overpricing or underpricing)

Engineering Contradiction:

Improvedata transfer speedVSAvoiddataset value assessment accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary sufficiency evaluation and generates dataset sufficiency-indicators before the actual data transfer and pricing decision. The AI program analyzes dataset characteristics, compares them against sufficiency parameters from past transfers, and produces valuation indicators in advance, allowing buyers and sellers to make informed decisions without delaying the transfer process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary AI program that acts as a mediator between data holders and buyers. This AI program objectively evaluates dataset sufficiency using trained models and sufficiency parameters, providing an independent assessment that helps both parties agree on fair pricing without prolonged negotiation or uncertainty.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If users obtain more training data to improve model accuracy, then model prediction accuracy increases, but data acquisition costs and complexity increase

Engineering Contradiction:

Improvemodel prediction accuracyVSAvoiddata acquisition complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system replaces manual data sufficiency assessment with an automated AI program that uses machine learning models to evaluate datasets. Instead of users manually analyzing dataset quality or negotiating based on intuition, the AI program automatically processes dataset characteristics, compares them against trained sufficiency parameters, and generates objective valuation indicators, simplifying the data acquisition process.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system transforms the complex problem of data sufficiency assessment into measurable parameters that can be processed by the AI program. Dataset characteristics are converted into quantifiable sufficiency parameters (such as data volume, diversity, quality metrics) that the AI model can analyze systematically, enabling objective comparison and valuation without requiring users to understand complex data science concepts.

Inventive Principle:

Principle #35Parameter changes

3Quantity of substance

If data holders charge higher prices for datasets, then potential revenue increases, but buyers may reject the data transfer

Engineering Contradiction:

Improvedataset priceVSAvoiddata transfer feasibility

Core Design Contradiction:

Quantity of substanceVSAdaptability or versatility

Solution Approach 1:

The system implements feedback through the AI program's evaluation mechanism, which analyzes dataset characteristics and compares them against sufficiency parameters from past successful transfers. This feedback loop provides objective information about what price range is appropriate for the dataset's actual sufficiency, preventing both overpricing (which would reject transfers) and underpricing (which would reduce revenue).

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20230359704A1Centralized repository and data sharing hub for establishing model sufficiency

Publication Date: 2023.11.09 TRUIST BANK

AI summary

A system for centralized transfer of model input data includes a computer to execute instructions. One instruction is to receive a dataset, with an unknown actual sufficiency, and including data suitable to model behavior and for representation on a sufficiency listing. One instruction is to modify the sufficiency listing to include a representation of the dataset. Another instruction is to generate sufficiency parameters in a sufficiency database utilizing a previous transfer module that interfaces with a previous version of the sufficiency listing. A further instruction is to use the dataset, the sufficiency parameters, and an artificial intelligence program to generate a dataset sufficiency-indicator and to connect the dataset sufficiency-indicator to the dataset. An additional instruction is to communicate to a user the dataset sufficiency-indicator and the representation of the dataset in association with the dataset in or to allow the user to estimate the actual sufficiency of the dataset.