Cooccurrence Dictionary Creation Using Clause-Level Semantic Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional cooccurrence dictionary creating systems generate low-quality dictionaries with many cooccurrences lacking semantic relationships and require large storage capacity due to the inclusion of irrelevant cooccurrences, especially when dealing with complex expressions.
Innovation Solution
A cooccurrence dictionary creating system that limits cooccurrence relationships to modification relationships within clauses, using a clause as the unit of analysis to extract only meaningful complex expressions, thereby reducing storage needs and focusing on semantic relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all cooccurrences within a predetermined range (document, paragraph, or sentence) are collected, then the quantity of cooccurrence data increases, but the quality of semantic relationships deteriorates due to inclusion of irrelevant cooccurrences
Solution Approach 1:
The patent extracts only the meaningful subset of cooccurrences that satisfy specific structural conditions: (1) both words are nouns, (2) both words belong to the same clause, (3) at least one word is a subject or object of the clause, (4) the other word is modified by a modifier of the subject or object. This extraction principle filters out irrelevant cooccurrences while preserving semantically significant ones, resolving the contradiction between quantity and quality.
2Adaptability or versatility
If all word chains (complex expressions) are included in the cooccurrence dictionary, then the coverage of complex expressions improves, but the storage capacity requirement increases significantly
Solution Approach 1:
The patent extracts only meaningful complex expressions by applying semantic criteria: (1) the complex expression must contain at least one noun that is a subject or object of the clause, (2) the other words must be modifiers of the subject or object. This selective extraction reduces storage requirements while maintaining coverage of semantically important complex expressions.
Solution Approach 2:
The patent applies different quality standards to different parts of the cooccurrence data: core cooccurrences involving subjects and objects are preserved with high priority, while other cooccurrences are filtered out. This local quality approach ensures that the most semantically important relationships are retained while reducing overall data volume.
3Quantity of substance
If the predetermined range for obtaining cooccurrences is widened from sentence to paragraph to document, then the quantity of cooccurrences increases, but the precision of semantic relationships decreases due to inclusion of unrelated cooccurrences
Solution Approach 1:
The patent extracts cooccurrences based on clause-level structural relationships rather than document-level frequency. By requiring that both words belong to the same clause and have specific grammatical roles (subject, object, modifier), the system maintains high semantic precision even when analyzing large document corpora, effectively filtering out unrelated cooccurrences that would otherwise be included in wider-range analysis.
Data Source
AI summary
A cooccurrence dictionary creating system includes: a language analyzing section which subjects a text to a morpheme analysis, a clause specification, and a modification relationship analysis between clauses, a cooccurrence relationship collecting section which collects cooccurrences of nouns in each clause of the text, modification relationships of nouns and declinable words, and modification relationships between declinable words as cooccurrence relationships, a cooccurrence score calculating section which calculates a cooccurrence score of the cooccurrence relationship based on a frequency of the collected cooccurrence relationship, and a cooccurrence dictionary storage section which stores a cooccurrence dictionary in which a correspondence between the calculated cooccurrence score and the cooccurrence relationship is described.