Automatic Sample Categorization via Token Pattern Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The lack of uniformity in sample labeling across different laboratories and workers leads to difficulties in automatically categorizing and accurately interpreting data for samples, particularly in medical and research contexts, where variations in naming conventions and notation styles hinder efficient processing and analysis.
Innovation Solution
A computer-implemented method and apparatus for automatic categorization of samples, which involves receiving data units, parsing them into ordered lists of specific tokens, merging similar lists to define common patterns, and categorizing samples based on these patterns, using techniques such as regular expressions to standardize labeling and reduce duplicity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual labeling with varied naming conventions is used, then sample identification flexibility is maintained, but automatic categorization accuracy deteriorates
Solution Approach 1:
The system transforms sample labeling data from varied textual formats into a standardized parameter structure through parsing into ordered lists of tokens. This parameter transformation enables automated processing while preserving the original labeling flexibility, as the system can handle multiple naming conventions by converting them to a common parameter representation for accurate categorization.
Solution Approach 2:
The patent introduces an intermediary processing layer that includes parsing, merging, and pattern matching components. This intermediary system acts as a bridge between diverse manual labeling conventions and automated categorization requirements, translating varied label formats into standardized patterns that enable accurate automatic categorization without restricting original labeling flexibility.
2Ease of operation
If diverse labeling conventions are accepted, then ease of sample collection is improved, but data processing complexity increases
Solution Approach 1:
The system segments the data processing task into distinct modular components: parsing individual labels into token lists, merging similar patterns across multiple labels, and matching against categorized patterns. This segmentation reduces overall processing complexity by breaking down the complex task of handling diverse labeling conventions into manageable, independent steps that can be processed systematically.
Solution Approach 2:
The system changes the parameter representation of labeling data from diverse textual formats to a standardized structured format (ordered lists of tokens). This parameter transformation simplifies subsequent processing operations by converting unstructured diverse labels into uniform parameters that are easier to compare, merge, and categorize automatically.
3Stability of the object's composition
If pattern merging is performed to standardize labels, then labeling uniformity is improved, but processing time for pattern recognition increases
Solution Approach 1:
The system performs preliminary pattern merging during the data collection phase, consolidating similar labeling patterns into standardized forms before the actual categorization task. This preliminary action creates a pre-processed set of unified patterns that reduces the computational burden during pattern recognition, thereby maintaining labeling uniformity while minimizing processing time during actual categorization operations.
Data Source
AI summary
A computer implemented method for automatic categorization of samples, the method comprising steps a computer is programmed to perform, the steps comprising: on a computer, receiving a plurality of data units, each one of the data units labeling a respective sample of a physical substance, parsing each one of the data units into at least one respective ordered list of one or more specific tokens, for at least one subgroup of the ordered lists, the subgroup consisting of ordered lists of a same number of specific tokens, merging at least two ordered lists in which each pair of specific tokens of a same order of in-ordered-list positioning are at least one of a group consisting of: a pair of identical specific tokens, and a pair of non-identical specific tokens matching a same general token, into an ordered list defining a pattern common to the ordered lists being merged, and categorizing a sample by matching a data unit labeling the sample being categorized, to at least one of a plurality of labeling patterns, each one of the labeling patterns being defined by a respective one of the ordered lists resultant upon the merging.