Event Detection via Graph Ontology Similarity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing event detection algorithms face challenges in efficiently identifying new events from massive data without requiring significant updates, leading to increased technology updating or upgrading costs.
Innovation Solution
The proposed method and system utilize an event detection approach that involves extracting instance data from text using an extraction model, constructing an extracted graph, and comparing it with ontology definition graphs of candidate events to determine similarities, thereby identifying events with reduced costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing event detection algorithms are used to identify new events, then event detection capability is improved, but technology updating or upgrading costs increase
Solution Approach 1:
The event detection system is segmented into two independent parts: a pre-built extraction model for entity and relationship extraction, and a dynamic graph ontology structure for event pattern matching. This segmentation allows the extraction model to remain stable while the graph ontology can be updated independently to accommodate new events, reducing overall system updating costs.
Solution Approach 2:
The graph ontology structure is designed to be dynamic and adaptable to new events. Instead of requiring complete retraining of the extraction model, the system dynamically updates the graph ontology with new event patterns and relationships, allowing the system to adapt to new events while maintaining the stability of the extraction model.
2Adaptability or versatility
If existing event detection algorithms are updated for new events, then adaptability to new events is improved, but technology updating costs increase
Solution Approach 1:
The graph ontology structure serves multiple functions: it stores event patterns, enables similarity matching between extracted graphs and known events, and facilitates efficient retrieval of relevant events. This multi-functionality reduces the need for separate components for each function, thereby reducing updating costs when adding new event types.
Solution Approach 2:
Instead of retraining the extraction model for new events, the system creates a copy of the extraction model's output (the extracted graph) and compares it against a database of known event patterns stored in the graph ontology. This copying approach allows new events to be handled by matching against existing patterns rather than requiring model retraining.
3Measurement precision
If event detection algorithms are updated frequently, then detection accuracy for new events is improved, but system complexity and maintenance costs increase
Solution Approach 1:
The system replaces the mechanical process of retraining extraction models with a computational approach using graph similarity matching. Instead of mechanically retraining the entire model, the system substitutes this with a more efficient process of extracting graphs and matching them against pre-stored event patterns using similarity algorithms, thereby maintaining accuracy while reducing complexity.
Data Source
AI summary
This specification discloses an event detection method and system. The method includes: obtaining a to-be-processed text; extracting one or more groups of instance data from the to-be-processed text by using an extraction model; determining one or more extracted triplets based on the one or more groups of instance data, to obtain an extracted graph; obtaining graph ontology definition data of one or more candidate events, and obtaining an ontology definition graph corresponding to each candidate event based on the graph ontology definition data; determining a similarity between the extracted graph and each ontology definition graph of the one or more candidate events; and determining an event corresponding to the to-be-processed text from the one or more candidate events based on each similarity.