基于图神经网络的人工自然语言语义挖掘仿真
Artificial Natural Language Semantic Mining Simulation Based on Graph Neural Network
周显春 1喻佳2
作者信息
- 1. 三亚学院信息与智能工程学院,海南 三亚 572022;三亚学院容淳铭院士工作站,海南 三亚 572022
- 2. 华东交通大学信息工程学院,江西 南昌 330013
- 折叠
摘要
语义挖掘工具可从批量非结构化人工自然语言文本数据中准确提取有用信息,但是由于网络环境文本具备半结构化、多尺度、海量、复杂关联等属性,导致文本数据通常维度较高,且仅有小部分节点存在明确标签,因此语义挖掘难度较大.提出基于图神经网络的人工自然语言语义挖掘方法.结合多头注意力机制和半监督图卷积神经网络对人工自然语言文本降维处理.联合改进的模糊C均值聚类算法和免疫单亲遗传算法,构建人工自然语言语义挖掘算法.实验结果表明,研究方法的聚类纯度、准确率和召回率均高于 95%,说明上述方法的应用性能较优.
Abstract
Semantic mining tools can accurately extract useful information from bulk data of unstructured artificial natural language text.However,due to the semi-structured,multi-scale,massive,and complex association attributes of network environment texts,text data usually has high dimensions and only a small number of nodes have clear la-bels,making semantic mining difficult.In this article,a method of mining artificial natural language semantics based on graph neural network was proposed.Firstly,multi-head attention was combined with semi-supervised graph convo-lution neural network to reduce the dimension of artificial natural language text.Then,the improved fuzzy c-means clustering algorithm was combined with a partheno-genetic algorithm based on immune mechanism to construct an ar-tificial natural language semantic mining algorithm.Experimental results show that the clustering purity,accuracy and recall rate of the proposed method are higher than 95%,proving its application performance.
关键词
图神经网络/人工自然语言/语义挖掘/多头注意力机制Key words
Graph neural network(GNN)/Artificial natural language/Semantic mining/Multi-head attention引用本文复制引用
基金项目
海南省自然科学基金(620MS064)
三亚市院地科技合作项目(2019YD26)
出版年
2024
NETL
NSTL
维普
万方数据