命名实体识别方法、装置、设备及存储介质
By employing a multi-layered processing model for named entity recognition, the problem of inconsistent labeling of multilingual search terms was solved, enabling accurate identification of named entities in financial texts and improving the efficiency of financial services and customer satisfaction.
CN117574899BActive Publication Date: 2026-07-17PING AN TECH (BEIJING) CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PING AN TECH (BEIJING) CO LTD
- Filing Date
- 2023-11-14
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In financial product scenarios involving multilingual search terms, existing technologies struggle to achieve consistent and unified labeling across multiple languages, resulting in insufficient accuracy in named entity recognition.
Method used
A named entity recognition model is adopted, including a semantic recognition layer, a spatial structure extraction layer, a feature fusion layer, and an output layer. Through semantic recognition processing, spatial structure feature extraction, and feature fusion, named entities in financial texts are accurately identified.
Benefits of technology
It improves the accuracy and efficiency of named entity recognition, thereby enhancing the efficiency of financial business processing and customer satisfaction.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117574899B_ABST
Abstract
本申请提供一种命名实体识别方法、装置、设备及存储介质，属于人工智能领域该方法包括：获取待识别命名实体的目标文本，并获取命名实体识别模型；将目标文本输入至语义识别层进行语义识别处理，得到目标文本的语义特征向量；将目标文本输入至空间结构提取层进行空间结构特征提取处理，得到目标文本的空间结构特征向量；将所述语义特征向量和空间结构特征向量输入至特征融合层进行特征融合处理，得到目标特征向量；将目标特征向量输入至输出层进行命名实体的预测，得到目标文本对应的命名实体。本方案能够准确且高效地识别出金融文本文件中的命名实体。本申请还涉及区块链技术，该命名实体识别模型可存储至该区块链。
Need to check novelty before this filing date? Find Prior Art