Knowledge Extraction Thresholds for Dynamic Concept Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge extraction techniques for stock market information are limited by the need for pre-association of target words with news, failure to associate non-similar issues, restricted grouping ranges, undisclosed utilization and update methods, and reliance on a single information source for certainty factor evaluation.
Innovation Solution
A knowledge extracting apparatus that receives electronic documents, extracts knowledge information using clue words and target words, forms associations, and updates certainty factors based on new information, allowing for dynamic grouping and evaluation of share price fluctuations without pre-defined associations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If knowledge extraction target words are associated with news information in advance, then knowledge information can be extracted efficiently, but the system cannot identify non-similar issues and lacks adaptability to new issues
Solution Approach 1:
The system dynamically updates the association between knowledge extraction target words and news information based on share price fluctuations. Instead of static pre-association, the system adapts associations in real-time by monitoring stock market data and automatically linking relevant news to appropriate target words when similarity is detected, thereby maintaining both extraction efficiency and adaptability to new issues.
Solution Approach 2:
The system performs self-updating of knowledge associations by automatically detecting share price fluctuations and autonomously establishing links between news information and knowledge extraction target words without requiring manual intervention. This self-service mechanism enables the system to adapt to new issues while maintaining efficient extraction operations.
2Speed
If issues are set in advance with explicit associations, then the system can quickly retrieve related information, but the range for grouping issues is limited
Solution Approach 1:
The system transforms static issue associations into dynamic ones by continuously monitoring share price fluctuations and automatically expanding the range of grouped issues. When new issues exhibit similar price movement patterns, the system dynamically incorporates them into existing issue groups, thereby maintaining fast retrieval speeds while expanding adaptability to new and diverse issues.
3Loss of information
If keyword extraction is performed based on stored news, then knowledge information can be obtained, but the method and update mechanism are not disclosed
Solution Approach 1:
The system implements a feedback mechanism where share price fluctuation data is fed back into the keyword extraction and association process. This feedback loop enables automatic updating of knowledge associations by comparing current market data with historical patterns, ensuring knowledge information remains current while maintaining a manageable extraction mechanism through automated decision rules.
4Measurement precision
If only prediction coefficients are updated based on share price fluctuations, then the system can maintain evaluation accuracy, but other knowledge information remains stale
Solution Approach 1:
The system extends the update mechanism from merely refreshing prediction coefficients to continuously updating all associated knowledge information including news associations, issue groupings, and target word linkages. This continuous update process ensures that all knowledge elements remain timely and accurate, not just the evaluation metrics.
Data Source
AI summary
Provided is a knowledge extracting apparatus for extracting knowledge information related to a knowledge-extraction target from an electronic document distributed continually in a state where the electronic document is not associated with the knowledge-information extraction target. A knowledge extracting apparatus according to one embodiment is a knowledge extracting apparatus including: an information receiving section for receiving an electronic document; a knowledge extracting section for extracting a concept from the electronic document based on a target word to extract knowledge information and a clue word to extract knowledge information and forming knowledge information in which the concept thus extracted and the target word are associated with each other; a storage section for storing the knowledge information thus extracted; and an information analysis section for, after the knowledge information is stored, analyzing the electronic document based on the knowledge information in the storage section.