AI Search Apparatus for Image Data Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search systems, such as those described in patent document 1, face difficulties in applying conversational scenarios to image data searches due to the high similarity and overlapping nature of image data, making it challenging to generate an optimal search criterion efficiently.
Innovation Solution
A search apparatus and method that includes a search target extraction unit, a score calculation unit, a question generation unit, and a search criterion generation unit, which reference a dataset of image and attribute information pairs, use a knowledge base to calculate scores, and update the search criterion based on user responses to questions, to efficiently narrow down search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a search criterion is generated based on user input alone, then the search system is simple to operate, but the search accuracy is low due to inability to grasp enormous amount of data
Solution Approach 1:
The patent introduces an AI assistant as an intermediary between the user and the search system. The AI assistant automatically generates search criteria by analyzing user intent and the enormous amount of data, resolving the contradiction by providing high search accuracy through complex AI processing while maintaining simple user operation through natural language queries.
Solution Approach 2:
The system enables self-service by allowing the AI assistant to autonomously generate and refine search criteria without requiring users to manually analyze data or construct complex search queries. The AI assistant automatically understands user intent and performs the complex task of criterion generation, achieving both high accuracy and ease of operation.
2Productivity
If manual search criterion generation is used, then the system requires less computational resources, but substantial effort and great amount of time are required to obtain target search results
Solution Approach 1:
The patent applies preliminary action by having the AI assistant pre-process and analyze the enormous amount of data before the actual search is executed. The AI assistant generates optimized search criteria in advance based on user intent, which reduces the computational burden during the actual search execution while maintaining high productivity.
Solution Approach 2:
The system dynamically changes search parameters by having the AI assistant analyze user intent and automatically adjust search criteria based on the enormous amount of data. This allows the system to achieve high search efficiency by optimizing parameters automatically, while computational resources are consumed intelligently rather than uniformly.
3Measurement precision
If image data with many similarities is searched using conventional methods, then the search covers broad ranges, but it is difficult to narrow down results efficiently
Solution Approach 1:
The patent implements feedback mechanisms where the AI assistant continuously refines search criteria based on search results and user responses. When searching image data with many similarities, the system analyzes the results, identifies differentiation opportunities, and automatically adjusts criteria to narrow down results efficiently, reducing the time loss while improving result differentiation.
Solution Approach 2:
The system applies dynamics by making search criteria adaptive and changeable based on real-time analysis of search results. The AI assistant dynamically adjusts search parameters when dealing with similar image data, allowing the system to efficiently differentiate and narrow down results rather than using static search criteria that would require manual refinement.
Data Source
AI summary
A search apparatus including: a search target extraction unit that references a data set use of a search criterion that includes attribute information, and extracts search target information that include the attribute information that match the attribute information of the search criterion; a score calculation unit that, in a case where the number of the extracted search target information pieces is not within a preset search result range, references a knowledge base and calculates a score for each of the attribute information pieces included in the extracted search target information pieces with use of a score function that has been determined in advance; a question generation unit that selects an attribute information piece based on the calculated scores, and generates question information use of the selected attribute information piece; and a search criterion generation unit that reflects an attribute information piece indicated by a response of the user to the question information in the search criterion, thereby generates a new search criterion.