Label-Table Information Extraction to Prevent NLP Error Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information category extraction methods in natural language processing rely on character extraction, leading to errors being propagated and inaccurate information category extraction.
Innovation Solution
A label table is established with row and column characters aligned to the order of a sample sentence, marked with information category labels, and used to train a model to extract target sentence information, eliminating the need for separate character and category extraction models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a pipeline method with separate character extraction model and category extraction model is used, then the extraction process can be divided into distinct steps, but errors in character extraction are propagated to category extraction leading to reduced accuracy
Solution Approach 1:
The patent combines character extraction and category extraction into a single unified model. The model takes the input sentence and directly outputs the category extraction result without relying on intermediate character extraction, thereby eliminating error propagation while maintaining a streamlined architecture.
2Ease of operation
If character extraction is performed as a preliminary step before category extraction, then the extraction process follows a clear sequence, but the dependency on character extraction results causes error propagation
Solution Approach 1:
The patent extracts and removes the intermediate character extraction step from the processing pipeline. The unified model directly processes the input sentence to produce category extraction results, eliminating the dependency on character extraction and preventing error propagation while maintaining operational simplicity.
3Adaptability or versatility
If two separate models are established for character extraction and category extraction, then each model can be optimized independently, but the overall system requires more computational resources and time
Solution Approach 1:
The patent merges character extraction and category extraction into a single unified model, eliminating the need for two separate models. This reduces computational resources and processing time while maintaining the ability to optimize the extraction process, as the unified model directly transforms input sentences to category extraction results without intermediate steps.
Data Source
AI summary
The present disclosure relates to an information obtaining method and apparatus, a device, and a medium. The method includes setting a label table corresponding to each sample sentence in a sample set, wherein row characters and column characters in the label table are set identically in accordance with an order of characters of the corresponding sample sentence; and marking cells composed of the row characters and the column characters in the label table with corresponding information category labels; taking each sample sentence in the sample set as input information to a model to be trained and the label table corresponding to each sample sentence as output information of the model to be trained, and performing model training according to a preset target function; and generating an information extraction model based on parameters of the trained model to extract target sentence information by the information extraction model.