Legal Knowledge Graph Construction via Joint Entity-Relation Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge graph construction methods in the legal field suffer from accuracy issues due to the pipeline approach, where entity extraction errors propagate and affect relation extraction, leading to inaccuracies in constructing domain knowledge graphs.
Innovation Solution
A method and device for constructing a legal knowledge graph using joint entity and relation extraction, employing a model architecture with a Bert pre-training encoder, BiLSTM binary classifiers, and inter-sentence relation determination to extract triples directly from unstructured legal texts, avoiding error propagation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional pipeline method is used for entity extraction and relation extraction, then the construction process is simple and sequential, but entity extraction errors propagate to relation extraction, reducing accuracy
Solution Approach 1:
The patent combines entity extraction and relation extraction into a unified joint extraction model that processes both tasks simultaneously. The model architecture integrates entity recognition and relation classification layers, allowing the system to extract entities and their relationships from legal texts in a single pass, thereby preventing error propagation while maintaining construction efficiency.
2Measurement precision
If joint entity and relation extraction is used, then extraction accuracy is improved by avoiding error propagation, but the model complexity increases
Solution Approach 1:
The joint extraction model is segmented into distinct functional layers: an encoding layer for text representation, an entity extraction layer for identifying entities, and a relation extraction layer for determining relationships. This segmentation allows the complex joint extraction task to be broken down into manageable components, reducing overall model complexity while maintaining high extraction accuracy.
3Loss of energy
If pipeline method is used, then computational resources are saved by processing tasks sequentially, but error propagation reduces knowledge graph quality
Solution Approach 1:
The joint extraction model maintains continuous processing of entity and relation extraction in a single unified framework. By processing both extraction tasks simultaneously rather than sequentially, the model avoids the interruption and error propagation inherent in pipeline methods, ensuring continuous improvement of knowledge graph quality without excessive computational resource consumption.
Data Source
AI summary
A method and device for constructing a legal knowledge graph based on joint entity and relation extraction. The construction method comprises the following steps: constructing a triple data set; design of a model architecture and training of a model, wherein the model architecture comprises an encoding layer, a head entity extraction layer and a relation-tail entity extraction layer; determination of the relation between the sentences of the text; triple combination and graph visualization. The design of the model framework of the present disclosure adopts a Chinese Bert pre-training model as an encoder. In the entity extraction part, two BiLSTM binary classifiers are used to identify the start position and end position of an entity. The head entity is first extracted, and then the tail entity corresponding to the entity relation is extracted from the extracted head entity.