Program-Guided Fact Checking for Explainable Complex Claims
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies have difficulty in addressing the inefficiencies of existing technologies in verifying the veracity of complex claims, particularly in the context of the veracity of complex claims that require multi-step reasoning, lack explainability, and require large amounts of training data.
Innovation Solution
A system and method that decomposes complex claims into simpler sub-tasks using a shared library of specialized functions, guided by a reasoning program generated by a large language model, and executes these sub-tasks to provide clear explanations and veracity labels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If automated fact-checking systems use traditional methods to verify complex claims, then they can process claims automatically, but they lack explainability and require large amounts of training data
Solution Approach 1:
The patent decomposes complex claims into multiple sub-claims and verifies each sub-claim separately using specialized functions. This segmentation approach provides clear explanations for each verification step while reducing the overall system complexity by breaking down the verification process into manageable components that can be handled independently.
Solution Approach 2:
The patent introduces a program generation module that creates verification programs as intermediaries between the complex claim and the verification process. These programs serve as explainable intermediaries that detail the verification steps, providing transparency without requiring the entire system to be fundamentally more complex.
2Productivity
If existing fact-checking systems handle complex claims requiring multi-step reasoning, then they can verify such claims, but they are inefficient and lack clarity in their reasoning process
Solution Approach 1:
The patent generates verification programs in advance that outline the multi-step reasoning process before actual verification occurs. This preliminary structuring of the verification approach improves efficiency by providing a clear roadmap, while simultaneously preserving reasoning explainability through the documented verification steps.
Solution Approach 2:
The verification process is divided into discrete sub-tasks, each handled by specialized functions. This segmentation improves productivity by allowing parallel processing of independent sub-claims while maintaining reasoning explainability through clear documentation of each verification step's contribution to the overall conclusion.
3Adaptability or versatility
If fact-checking systems use shared libraries of specialized functions, then they can handle various fact-checking scenarios flexibly, but they require sophisticated program generation and execution mechanisms
Solution Approach 1:
The patent creates a shared library of specialized verification functions that can handle multiple fact-checking scenarios. This universal approach increases adaptability and versatility, as the same framework can verify different types of claims using appropriate specialized functions, while the program generation module automatically selects and orchestrates the right functions for each claim type.
Solution Approach 2:
The system uses program generation that automatically selects and orchestrates appropriate specialized functions based on the claim being verified. This self-service approach increases versatility without proportionally increasing complexity, as the system autonomously determines the verification pathway without requiring manual configuration for each new claim type.
Data Source
AI summary
A method and system are arranged for fact checking electronically posted claims. The system includes an input/output device for inputting a textual complex claim as a natural language statement with a combination of two or more facts and outputting an indication that the textual complex claim is true or false. A processor connected to the input/output device generates, by a program generation module, a reasoning program for the input complex claim as a sequence of reasoning steps, where each step directs to a function in a set of sub-task functions. The processor generates, by a program execution module, each step in the reasoning program to a respective sub-task function and executes the respective sub-task function in order to provide the outputted indication.