Keyword Clustering for Stock Classification via Vector Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for clustering keywords related to stocks lack efficiency in identifying mutual associations and adapting to real-time market events, leading to suboptimal stock classification and investment decisions.
Innovation Solution
An electronic device performs a method of clustering keywords by identifying text sets, keyword sets, and vectors based on the degree of association between target keywords and other keywords, using techniques like named entity recognition and morpheme analysis, and generates information about clustered sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual classification of stocks is used, then simplicity and ease of operation are maintained, but productivity and measurement precision of stock associations are insufficient
Solution Approach 1:
The patent segments the stock classification process into distinct functional modules: text data acquisition module, keyword extraction module (using NER and morpheme analysis), vector generation module, similarity calculation module, and clustering module. This segmentation enables automated processing while maintaining manageable system complexity through modular design.
Solution Approach 2:
The patent introduces vectors as an intermediary representation between keywords and similarity metrics. Instead of directly comparing keywords, the system converts keywords to vector representations, calculates similarity between vectors, and then performs clustering. This intermediary approach enables efficient automated processing while maintaining measurement precision.
2Adaptability or versatility
If traditional keyword clustering is used, then system simplicity is maintained, but adaptability to real-time market events and measurement precision of associations are insufficient
Solution Approach 1:
The patent implements a dynamic clustering system that continuously acquires new text data, extracts updated keywords, recalculates vector representations, and re-performs similarity calculations and clustering. This dynamic process enables the system to adapt to real-time market events while maintaining precise measurement of keyword associations through vector-based similarity metrics.
Solution Approach 2:
The system incorporates feedback mechanisms where clustering results and similarity measurements inform subsequent keyword extraction and vector generation processes. The system continuously refines its understanding of stock associations by feeding back clustering outcomes to adjust keyword weighting and vector representation, thereby improving both adaptability and measurement precision over time.
3Measurement precision
If automated text analysis with NER and morpheme analysis is used, then measurement precision of keyword extraction is improved, but device complexity and processing time increase
Solution Approach 1:
The patent performs preliminary text preprocessing including segmentation into text elements and initial keyword identification before applying NER and morpheme analysis. By preparing the text data in advance with basic filtering and structuring, the system reduces the computational burden of subsequent complex analyses, thereby maintaining high extraction precision while reducing overall processing time.
4Measurement precision
If vector-based similarity calculation is used, then measurement precision of keyword association is improved, but computational complexity and energy consumption increase
Solution Approach 1:
The patent transforms keywords into vector representations with specific dimensional parameters, where each dimension corresponds to a particular keyword or feature. By changing the representation parameters from discrete keywords to continuous vector spaces, the system achieves precise similarity measurement through mathematical operations. The system optimizes energy consumption by selecting appropriate vector dimensions and using efficient similarity calculation algorithms.
Data Source
AI summary
A method of clustering keywords includes identifying a text set including at least one text element, identifying a keyword set including a keyword in the at least one text element, the keyword set including at least one target keyword, identifying at least one vector corresponding to each of the at least one target keyword. An element of the at least one vector is identified based on a degree of association between the at least one target keyword and each of keywords included in the keyword set, the degree of association is identified based on the text set. The method includes identifying a similarity between the at least one target keyword, based on the identified at least one vector. The method includes identifying at least one set including at least some of the at least one target keyword, by clustering the at least one target keyword based on the similarity.