基于深度神经网络的软件机器人语义识别方法

Semantic Recognition Method of Software Robot Based on Deep Neural Network

由于软件机器人语义识别中存在噪音,易产生表述不清等问题,增加了数据分析难度,影响了语义识别效果,为此,设计一种基于深度神经网络的软件机器人语义识别方法。通过Python语言中的urllib模块和NPL模块抓取二分类语料中的正负语料,搭建软件语料库。利用输出模块、解码器模块、编码器模块以及数据输入模块构建深度卷积神经网络模型,对融合后的张量进行处理后,获取语义分割结果。设计联合语义智能机器人识别模型,通过合并意图识别与实体识别,实现软件机器人语义识别。测试结果表明,设计方法的平均F1值高于0.82,最高错误率低于0.592%,语义分割标准度量值高于0.83,对大量数据进行语义识别时花费时间较短,证明了设计方法具有较好的语义识别效果。

The noise and unclear expression in the semantic recognition of software robot increase the difficulty of data analysis and affect the effect of semantic recognition,hence,a semantic recognition method of software robot based on deep neural network is designed.Through the urllib module and NPL module in Python language,the positive and negative corpora in the binary classification corpus are captured,and the software corpus is built.The deep convolution neural network model is constructed by using the output module,decoder module,encoder module and data input module.After processing the fused tensor,the semantic segmentation results are obtained.We design a joint semantic intelligent robot recognition model,and realize the semantic recognition of software robot by combining intention recognition and entity recognition.The test results show that the average F 1 value of the design method is higher than 0.82,the maximum error rate is lower than 0.592%,and the semantic segmentation standard measurement value is higher than 0.83.It takes a short time to perform semantic recognition on a large number of data,which proves that the design method has a good semantic recognition effect.