Event Recognition Model Using Trigger and Context Words
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current news event recognition methods are inefficient and lack portability, struggling to accurately recognize diverse news events due to their reliance on mode matching recognition, which is slow and ineffective in handling the high frequency of news updates.
Innovation Solution
An event recognition method and model training approach that utilizes trigger words and context words to classify events, employing fully-connected neural networks and attention weights to improve recognition efficiency and accuracy, with a focus on anonymizing named entities to enhance generalization and reduce the impact of specific information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If mode matching recognition method is used, then event recognition can be performed, but recognition efficiency is low and portability is poor
Solution Approach 1:
The patent replaces the mechanical mode matching recognition method with a neural network-based event recognition model. The model uses trigger words and context words to automatically identify event types, substituting the manual or rule-based matching process with an intelligent system that processes events more efficiently and accurately.
Solution Approach 2:
The patent changes the recognition approach from exact mode matching to probabilistic neural network classification. By using trigger words and context words as input features, the system transforms the recognition problem into a parameter-based classification task that can handle diverse event types more effectively.
2Measurement precision
If trigger word and context word based recognition is used, then recognition accuracy is improved, but model complexity increases
Solution Approach 1:
The patent segments the event recognition process into distinct components: trigger word identification, context word extraction, and classification. By dividing the input sentence into these functional parts, the model can process each element separately and combine their contributions, improving accuracy while maintaining manageable complexity through modular processing.
Solution Approach 2:
The patent introduces context words as intermediaries between the trigger word and the event type classification. These context words provide additional semantic information that helps disambiguate event types, acting as a bridge that enriches the recognition process without directly increasing model structural complexity.
3Adaptability or versatility
If named entities are anonymized, then model generalization is enhanced, but information loss occurs
Solution Approach 1:
The patent extracts and removes specific named entity information from the input, replacing it with anonymized placeholders. This extraction process retains the structural and semantic relationships necessary for event recognition while eliminating entity-specific details that would reduce model generalization to new, unseen events.
Data Source
AI summary
This application discloses an event recognition method, including: obtaining, by a terminal device, a target sentence used for recognizing a type of a target event; processing, by the terminal device, a target sentence based on an event recognition model, to obtain the type of the target event, the event recognition model being used for determining the type of the target event by using a trigger word in the target sentence and at least one context word of the trigger word, the trigger word being used for indicating candidate types of the target event, and the candidate types including the type of the target event; and outputting, by the terminal device, the type of the target event. According to the technical solutions of this application, an event recognition process is performed by using a trigger word and a context word of the trigger word. Such a word classification method for event recognition can improve recognition efficiency, and reference to the context word can improve accuracy of the event recognition.