【摘要】 现阶段智能机器人的应用越来越广泛，许多电商商城都会配备智能机器人辅助来减少人工客服的工作量，同时也能省下来客户的大部分时间，针对智能机器人收到指令后意图识别不准确的问题，提出将RNN与BILSTM融合的深度学习模型。首先将文本进行处理过后，经过嵌入层转换为词向量，先经过RNN层，再将循环神经网络输出的结果输入BILSTM中，得到的结果在经过线性层后通过Softmax函数进行分类，最终经过多组实验验证并进行比对，实验结果表明，使用RNN-BILSTM融合模型相比较于单一的深度学习模型分别在召回率、精确率、准确率以及F1-Score上面提升了4.31%,3.64%、3.6%、1.5%，证明了模型融合在对机器人进行意图识别文本分类任务上的合理性以及有效性。更多还原

【Abstract】 At present,the application of intelligent robot becomes more and more widely,many commercial malls are equipped with intelligent robots to reduce artificial customer service work,it also can save most of the time for customers,for intelligent robot received instruction intention recognition inaccurate problem,deep learning model combined with RNN and BILSTM is proposed. First of all,the text is processed,after the embedding layer is converted into word vector,it first passes through the RNN layer,and then the output results of the cyclic neural network are input into the BILSTM. After passing through the linear layer,the obtained results are classified by the Softmax function. Finally,after the verification and comparison of several groups of experiments,the experimental results show that compared with the single deep learning model,the recall rate,precision rate and F1-score of the fusion model using RNN-BILSTM are respectively increased by 4.31%,3.64%,3.6% and 1.5%. It is proved that the model fusion is reasonable and effective for the robot’s intention recognition text classification task.更多还原