Abduction Apparatus Probability Truth Value Assignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abduction processing models, such as the Least-specific and Most-specific models, face inefficiencies in assigning truth values to query logical formulae, leading to inappropriate inference results and decreased processing efficiency.
Innovation Solution
An abduction apparatus and method that includes a probability calculation unit, a closed world assumption probability calculation unit, and a solution hypothesis determination unit to calculate probabilities and determine solution hypotheses without degrading processing efficiency, by assigning truth values to query logical formulae using observation and knowledge information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the Least-specific model is used to assign truth values, then the efficiency of abduction processing is improved, but the accuracy of inference results deteriorates because truth values are arbitrarily set to unknown
Solution Approach 1:
The patent changes the parameter of truth value assignment from binary (known/unknown) to a continuous probability scale. The probability calculation unit computes probabilities for each candidate hypothesis being an explanation, and the solution hypothesis determination unit selects based on these probability values, thereby resolving the contradiction between efficiency and accuracy.
2Measurement precision
If the Most-specific model is used to assign truth values, then the accuracy of inference results is improved, but the efficiency of abduction processing deteriorates due to constructing the Herbrand universe
Solution Approach 1:
The patent extracts only the necessary candidate hypotheses from the full Herbrand universe that are relevant to explaining the observation information. The candidate hypothesis generation unit generates a limited set of candidate hypotheses based on the query logical formula and background knowledge, avoiding the need to construct and evaluate the entire Herbrand universe, thus maintaining accuracy while improving efficiency.
3Adaptability or versatility
If candidate hypotheses are generated using query logical formula and background knowledge, then the completeness of hypothesis coverage is improved, but the complexity of evaluation increases
Solution Approach 1:
The patent segments the evaluation process into distinct functional units: the probability calculation unit evaluates each candidate hypothesis independently by calculating its probability of being an explanation, and the solution hypothesis determination unit aggregates these probabilities to determine the final solution. This segmentation simplifies the overall evaluation complexity while maintaining complete hypothesis coverage.
Data Source
AI summary
An abduction apparatus 1 includes: a probability calculation unit 2 configured to calculate, with respect to each of candidate hypotheses generated using observation information and knowledge information, a probability that the candidate hypothesis is an explanation regarding the observation information; a closed world assumption probability calculation unit 3 configured to calculate, with respect to the candidate hypotheses, a closed world assumption probability that the candidate hypothesis is an explanation regarding a first-order predicate logic literal to which a new truth value is determined as a result of assuming a closed world assumption; and a solution hypothesis determination unit 4 configured to determine a solution hypothesis that is a best explanation regarding the observation information from the candidate hypotheses using the probability and the closed world assumption probability.