文本分类方法、装置、设备及计算机可读存储介质
By combining bidirectional long short-term memory networks, multiple conditional random fields, and self-attention networks, the structural dependencies between adjacent words in text are captured, solving the problem of inaccurate identification of coherent viewpoint spans in existing text classification methods and improving the accuracy of text classification.
CN118673138BActive Publication Date: 2026-07-17BEIJING CO WHEELS TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING CO WHEELS TECH CO LTD
- Filing Date
- 2023-03-16
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing text classification methods rely on neural networks to learn contextual structure information, which leads to inaccurate recognition of coherent viewpoint spans and affects classification accuracy.
Method used
By introducing bidirectional long short-term memory networks, multiple conditional random fields, and self-attention networks, the structural dependencies between adjacent words in the text are captured, and global and local semantic information are combined to perform matrix concatenation to output the most likely text category.
Benefits of technology
It improves the accuracy of text classification, making the classification results closer to the intent of the text data and enhancing the ability to identify the span of coherent viewpoints.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN118673138B_ABST
Abstract
本公开涉及一种文本分类方法、装置、设备及计算机可读存储介质，该方法包括：获取文本数据的词向量文件；将词向量文件输入双向长短时记忆网络进行编码，输出文本数据的上下文信息；将文本数据的上下文信息输入多重条件随机场，得到文本数据的局部语义信息；将文本数据的上下文信息输入自注意力网络层，得到文本数据的全局语义信息；将文本数据的局部语义信息与文本数据的全局语义信息进行矩阵拼接，得到文本数据的融合语义信息；将融合语义信息输入预先训练好的文本分类模型，输出概率最大的文本类别作为文本数据的文本分类结果。本公开通过引入条件随机场来捕获文本中相邻单词间的结构依赖性，以提高文本分类的准确性。
Need to check novelty before this filing date? Find Prior Art