Recurrent Neural Network for Automated Text Expression Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual text checking for extracting and summarizing information from large texts is inefficient, requiring significant time and effort to identify and calculate relationships among entities.
Innovation Solution
A method and apparatus using a recurrent neural network to extract and process words from text, determining valid operators and generating expressions by inputting nodes and operators into sub-networks until no valid operators are found, thereby automating the extraction of expressions from text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual text checking is performed to extract and summarize expressions, then accuracy of expression extraction can be maintained, but time consumption increases significantly
Solution Approach 1:
The patent replaces manual mechanical text checking with an automated recurrent neural network system. The RNN processes text sequences to automatically extract expressions and calculate relationships between entities, substituting human cognitive processing with computational algorithms that achieve both accuracy and speed.
Solution Approach 2:
The system enables self-service by automatically performing expression extraction and relationship calculation without requiring human intervention. The RNN model processes text independently, identifying entities, extracting expressions, and verifying relationships autonomously based on learned patterns from training data.
2Reliability
If manual checking of relationships among entities is performed, then reliability of extracted information can be ensured, but productivity decreases
Solution Approach 1:
The patent substitutes manual relationship verification with an automated RNN-based system that calculates and validates relationships between entities. The model processes text to identify mathematical or logical relationships between numerical entities, providing reliable extraction while maintaining high productivity through automation.
3Productivity
If automated expression extraction is implemented, then productivity increases, but device complexity increases
Solution Approach 1:
The patent extracts and isolates the core functionality of expression extraction and relationship calculation into a dedicated RNN-based system. By separating this specific function from general text processing, the system achieves high productivity for expression extraction while managing complexity through modular design focused on the specific task at hand.
Data Source
AI summary
The present disclosure provides a method and an apparatus for obtaining an expression from characters. The method may include: extracting N words under test from a text under test in an arrangement order; inputting an i-th node in the first-level operation, each node of a first node to an i−1 th node in the first-level operation, and a predefined set of operators into a sub-network of a recurrent neural network to obtain nodes of a second-level operation; determining a valid operator in the first-level operation according to the nodes of the second-level operation; performing multi-level operations until the number of valid operators in a M-level operation is determined to be 0 according to the obtained nodes of the M+1-level operation; and generating the expression from the text under test according to valid operators in the first-level operation to the M−1-level operation and words corresponding to valid nodes.