Diagnostic Report Search Using Semantic Term Combinations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional diagnostic report searching methods are inefficient due to full-text search techniques, which require significant time and often fail to identify relevant reports written in different expressions, leading to decreased interpretation efficiency and the potential for missing crucial reference materials.
Innovation Solution
A diagnostic report searching apparatus that structures sentences into semantic units, generates combinations of related terms, counts their occurrences, and extracts related keywords to facilitate rapid and accurate retrieval of relevant reports.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full-text search is performed to find diagnostic reports, then all reports including the search term can be retrieved, but the search time becomes excessively long and interpretation efficiency decreases
Solution Approach 1:
The patent segments the diagnostic report text into structured elements (findings, diagnoses, procedures) and further divides findings into description units with semantic information. This segmentation allows the search to focus on specific structured fields rather than scanning entire documents, significantly reducing search time while maintaining completeness of relevant results.
Solution Approach 2:
The patent performs preliminary structuring and semantic analysis of diagnostic reports before the actual search occurs. By pre-processing the text into structured format with extracted semantic units during report creation or indexing, the system prepares the data in advance for rapid retrieval, eliminating the need for time-consuming full-text scanning during the search phase.
2Reliability
If full-text search is used to retrieve diagnostic reports, then a comprehensive list is generated, but it becomes difficult to determine which reports are actually useful as reference materials
Solution Approach 1:
The patent applies local quality by enhancing specific parts of the search results with structured semantic information. Instead of presenting uniform text snippets, the system highlights and organizes findings, diagnoses, and procedures with their semantic units, making it easy to quickly assess the relevance and quality of each report without reading the entire document.
Solution Approach 2:
The patent replaces the mechanical approach of manual full-text scanning with an automated semantic analysis system. By using computational methods to extract and compare semantic units from structured data, the system automatically identifies and ranks the most relevant reports, substituting human effort with intelligent automation.
3Device complexity
If simple character-string comparison is used for search, then the search process is simple, but diagnostic reports written in different expressions cannot be extracted even if they are actually related
Solution Approach 1:
The patent changes the parameter of search comparison from simple character-string matching to semantic unit comparison. By transforming text into structured semantic representations and comparing these standardized units, the system can identify equivalent meanings across different expressions while maintaining a manageable level of complexity through automated processing.
Solution Approach 2:
The patent introduces semantic units as an intermediary layer between the raw text and the search comparison process. This intermediary structure standardizes different expressions into comparable semantic representations, allowing the system to recognize equivalent meanings without requiring complex natural language understanding, thus bridging the gap between simplicity and accuracy.
Data Source
AI summary
According to embodiments, a diagnostic report search supporting apparatus and a diagnostic report searching apparatus each have a report registering part, a structuring processing part, a related-term analyzing part, a counting part, and a keyword extracting part. The structuring processing part extracts terms from a sentence written in a diagnostic report, and classifies the terms into predetermined kinds. The related-term analyzing part generates combinations each composed of two or more terms based on the plurality of terms having been extracted. The counting part counts the existence number of same combinations in the plurality of combinations, and extracts combinations whose existence numbers are a predetermined number or more. The keyword extracting part extracts a combination including a desired keyword, and extracts a term other than the desired keyword as a related keyword.