Morphological Analysis Model Generation for Contextual Word Meaning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semantic processing technologies struggle with word ambiguity and require extensive processing time due to dimensional compression of word vectors.
Innovation Solution
An information processing device that performs morphological analysis on character strings to generate a learned model using sub-word context matrices with predicate labels, enabling efficient inference of word meanings in varying contexts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If one word vector is assigned to one word for semantic processing, then the processing structure is simple, but the system cannot deal with word ambiguity
Solution Approach 1:
The patent segments the single word vector into multiple word vectors, where each vector corresponds to a different meaning or context of the word. This segmentation allows the system to represent multiple meanings simultaneously while maintaining a structured approach to semantic processing.
Solution Approach 2:
The patent introduces a new dimension to the word representation by adding context information and multiple meanings as additional dimensions. Instead of a single flat vector, the system uses multi-dimensional vectors that encode different semantic aspects, enabling the model to capture word ambiguity without significantly increasing structural complexity.
2Quantity of substance
If dimensional compression is applied to word vectors for learning, then the vector size is reduced, but the processing time increases enormously
Solution Approach 1:
The patent changes the parameters of the learning process by avoiding dimensional compression altogether. Instead of reducing vector dimensions through compression, the system maintains the original vector dimensions and uses alternative methods (such as attention mechanisms or context-based filtering) to achieve efficient processing, thereby reducing processing time while preserving vector information.
3Adaptability or versatility
If multiple word vectors are generated to represent different meanings, then word ambiguity is handled effectively, but the processing complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-processing the text to identify context information and potential multiple meanings before the main processing stage. This preliminary segmentation and context extraction reduce the complexity of subsequent processing by organizing the data in advance, making it easier to handle multiple word vectors efficiently.
Solution Approach 2:
The patent introduces an intermediary layer or mechanism that manages the multiple word vectors. This intermediary component (such as an attention mechanism or context-based selector) mediates between the multiple possible meanings and the final interpretation, reducing processing complexity by systematically selecting or combining the relevant vectors based on context.
Data Source
AI summary
An information processing device includes an acquisition unit that acquires a plurality of character strings, a morphological analysis execution unit that executes morphological analysis on the plurality of character strings, and a generation unit that generates a learned model for inferring meaning of a first word included in a plurality of words among a plurality of speech-parted words obtained by executing the morphological analysis on the plurality of character strings based on a plurality of words as a plurality of predicates among the plurality of speech-parted words and a plurality of words among the plurality of speech-parted words.