Event Extraction Model Using Sequential Question Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current event extraction methods have low accuracy in extracting structured information from text, leading to inefficient and inaccurate results in various fields such as finance and medicine.
Innovation Solution
An event extraction method that involves acquiring text information, determining question information ranked in sequential order, and inputting vector information, including answer mark and position vectors, into an extraction model to improve extraction accuracy and speed by considering the extraction information of previous questions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current event extraction methods are used, then the extraction process is simple, but the accuracy of structured information extraction is low
Solution Approach 1:
The extraction process is segmented into multiple sequential question information pieces, each targeting specific event arguments. The text is processed through multiple extraction rounds where each round answers a specific question, dividing the complex extraction task into manageable segments that improve accuracy without overwhelming complexity
Solution Approach 2:
Question information is predetermined and ranked in sequential order before extraction begins. Answer mark vectors are pre-calculated based on initial text marks, and the extraction model is prepared with predetermined question sequences, allowing the system to efficiently navigate the extraction process with pre-planned steps
2Measurement precision
If multiple question information pieces are processed sequentially, then the extraction accuracy improves, but the processing time increases
Solution Approach 1:
Question information is predetermined and ranked in sequential order before extraction begins. Answer mark vectors are pre-calculated based on initial text marks, reducing computation time during actual extraction. The systematic preparation of question sequences and mark vectors upfront minimizes processing time while maintaining high accuracy
Solution Approach 2:
The extraction process uses feedback from previous question answers to update answer mark vectors for subsequent questions. Extraction information from earlier questions informs and refines later extraction steps, creating an efficient feedback loop that improves accuracy without requiring complete re-processing of the text
3Productivity
If the search range for each question is kept wide, then the completeness of extraction is high, but the efficiency decreases
Solution Approach 1:
The answer mark vectors provide local quality indicators that highlight specific regions of text most relevant to each question. Instead of searching the entire text uniformly, the system focuses computational resources on locally identified relevant areas marked by the answer mark vectors, maintaining completeness while dramatically improving efficiency
Solution Approach 2:
The text search space is segmented into relevant and irrelevant regions based on answer mark vectors. Each question's extraction focuses on its specific relevant segment rather than the entire text, and the sequential question structure ensures that different segments are explored systematically without redundancy
Data Source
AI summary
The present disclosure provides an event extraction method, an event extraction device and an electronic device, and it relates to the field of computer data processing, in particular to the field of knowledge graph. The event extraction method includes: acquiring text information; determining a plurality of pieces of question information ranked in a sequential order in accordance with the text information; and inputting vector information for each piece of question information into an extraction model in accordance with the sequential order to acquire extraction information for each piece of question information.