基于融合自然语言处理的语义分析方法研究

Research on Semantics Analysis Method Based on Fusion Natural Language Processing Technology

面对电力复杂的数据环境下,传统单一的数据分析技术已经无法满足现实应用精准需求,需集多种策略优势为一体的模型综合处理数据信息。针对目前的发展瓶颈,采用融合性自然语言处理技术预测设备行为发展趋势,提取数据语义,通过小波分解法去除词义噪声,利用贝叶斯先验知识模型来推导后验概率,从而作为神经网络动态权值变化及语义预测分布的依据。通过实验测试证明了这种预测方法的可靠性及优越性。

In the complicated data environment, the traditional single data analysis technology can’t meet the demand of practicalapplications. It is necessary to integrate multiple strategic advantages into one model to process data information comprehensively.According to the current development bottleneck, the fusion natural language processing technology is used to predict the trend ofequipment behavior. Firstly, the data semantics are extracted and the meaning noise is removed by wavelet decomposition. Then theBayesian prior knowledge model is used to derive the posterior probability, which is used as a basis for dynamic weight change andsemantic prediction distribution in neural networks. The experimental test results prove the reliability and superiority of this method.