使用图扩散变换器进行方面级情感分类的方法和系统
By using a graph diffusion transformer model to perform multi-hop attention diffusion on the dependency tree graph, the problem of insufficient capture of syntactic relations of distant nodes in existing technologies is solved, thereby improving the accuracy of aspect-level sentiment classification.
CN116194912BActive Publication Date: 2026-07-17JINGDONG TECH HLDG CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JINGDONG TECH HLDG CO LTD
- Filing Date
- 2021-09-17
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing aspect-level sentiment classification methods struggle to effectively capture syntactic relationships between distant nodes, resulting in insufficient accuracy in sentiment classification.
Method used
The Graph Diffusion Transformer (GDT) model is adopted. Through multi-hop attention diffusion of the dependency tree graph, the attention matrix and graph attention diffusion of the dependency tree graph are calculated, the embedding of the dependency tree graph is updated, and the aspect terms are classified in combination with the classifier.
Benefits of technology
It improves the accuracy of aspect-level sentiment classification, effectively captures the syntactic relationships between distant nodes, and enhances the precision of sentiment classification.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116194912B_ABST
Abstract
一种方面级情感分类系统和方法。该系统包括计算设备，该计算设备包括处理器和存储有计算机可执行代码的存储设备。所述计算机可执行代码被配置为：接收具有被标记的方面术语和上下文的语句；将语句转换为依存树图；基于图中任意两个节点之间的一跳注意力，计算依存树图的注意力矩阵；根据注意力矩阵，计算任意两个节点的多头注意力扩散；使用多头扩散注意力获得图的更新嵌入；基于图的更新嵌入对方面术语进行分类，以获得方面术语的预测分类；基于方面术语的预测分类和地面真值标签，计算损失函数；以及基于损失函数调整计算机可执行代码中模型的参数。
Need to check novelty before this filing date? Find Prior Art