Information Processing Device Probability Distribution Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to identify relevant information that is actually relevant to a character string including a numerical value, especially when using a numerical range method.
Innovation Solution
An information processing device that acquires an input value as a numerical value and a feature word related to the input value, calculates probability distribution information, and determines similarity levels between the input value and target numerical values based on this information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a numerical range method is used to acquire information including a numerical value similar to the inputted numerical value, then information including similar numerical values can be acquired, but it is unclear which ones out of a plurality of pieces of information acquired are the relevant information actually relevant to the character string
Solution Approach 1:
The patent changes the parameter from simple numerical range matching to probability distribution-based similarity calculation. By representing numerical values as probability distributions and calculating similarity levels between distributions, the system achieves more precise identification of relevant information while maintaining the ability to handle multiple information pieces systematically
Solution Approach 2:
The patent replaces the mechanical numerical range comparison method with a probabilistic similarity calculation system. Instead of using fixed range thresholds, the system uses probability distribution functions to calculate similarity levels, enabling more accurate relevance determination among multiple information pieces
Data Source
AI summary
An information processing device includes an acquisition control unit that acquires an input value as a numerical value and a feature word as a word related to the input value, acquires probability distribution information as information regarding probability distribution and corresponding to the feature word, and acquires a plurality of target numerical values as a plurality of numerical values corresponding to the feature word and a calculation unit that calculates a plurality of similarity levels each being a level of similarity between the input value and each of the plurality of target numerical values based on the input value, the probability distribution information and the plurality of target numerical values.


