Case Search System for Notation Fluctuation Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional natural language processing methods fail to accurately handle notation fluctuations in technical terms across languages, leading to decreased learning accuracy and inability to effectively manage synonyms with different notations.
Innovation Solution
A case search method that preprocesses sentences by morphological analysis, unifies word groups with notation fluctuations into synonyms, and vectorizes the integrated sentences for learning, using a database that stores word vectors across languages, enabling effective handling of notation variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional natural language processing methods are used to vectorize words based on word distribution, then processing speed is maintained, but learning accuracy deteriorates due to inability to handle notation fluctuations in technical terms
Solution Approach 1:
The patent applies preliminary action by performing notation normalization and synonym unification before vectorization. The system pre-processes technical terms by identifying notation fluctuations (e.g., 'organise' vs. 'organize') and unifying them into canonical forms before the vectorization process, thereby improving learning accuracy without significantly increasing overall processing complexity.
Solution Approach 2:
The patent introduces an intermediary component (synonym dictionary/unification module) that mediates between raw text input and vectorization. This intermediary layer handles notation fluctuations by mapping variant forms to unified synonyms, allowing the core vectorization process to remain efficient while accurately handling technical term variations.
2Measurement precision
If words with notation fluctuations are treated as different words in conventional learning, then word diversity is maintained, but search accuracy deteriorates due to inability to recognize synonyms
Solution Approach 1:
The patent applies merging by combining multiple notation variants of the same technical term into a single unified representation. The system identifies words with notation fluctuations (e.g., 'sulphate' vs. 'sulfate') and merges them into unified synonyms, ensuring that search operations recognize all variants as equivalent, thereby improving search accuracy while maintaining adaptability to different notations.
Solution Approach 2:
The patent implements universality by creating a unified synonym representation that serves multiple notation variants simultaneously. The unified synonym acts as a universal key that represents all notation fluctuations of a technical term, enabling the system to handle diverse word forms with a single standardized representation, thus improving both search accuracy and notation adaptability.
3Measurement precision
If morphological analysis is applied to divide sentences into words, then word-level processing precision is improved, but processing time increases
Solution Approach 1:
The patent applies local quality by applying morphological analysis selectively to regions of the text where technical terms with notation fluctuations are likely to appear, rather than uniformly processing the entire text. This targeted approach maintains word-level processing precision for critical technical terms while reducing overall processing time by avoiding unnecessary fine-grained analysis in less critical regions.
Data Source
AI summary
In order to provide a case search system in which a search case corresponding to the notation fluctuation of a word within a sentence is a learning object, a sentence constituting a search target case is divided into a plurality of words by a morphological analysis, and a synonym integration unit 20 converts a predetermined word group within the sentence based on the relationship between the plurality of words into a single word consisting of another language in a synonym relationship with the word group. Further, the predetermined word group within the sentence is converted to a sentence with words stored in the corpus and the sentence is integrated, and the search target case composed of the integrated sentence is vectorized by leaning.