【摘要】 为提升仿人机器人人机交互体验,以序列到序列(SEQ2SEQ)神经网络模型为基础设计了SHFR-Ⅲ仿人机器人的对话系统,以集成学习形式构建适用于机器人交互的情感分析模型。情感分析模型设计过程中实验了多个模型分类效果,通过多模型的集成,提升了基于文本的情感分析的准确率。在对话系统设计中,以最大互信息作为模型的目标函数,对比实验表明最大互信息能有效减少SEQ2SEQ模型安全无意义回答的出现,提升人机交互质量。更多还原

【Abstract】 The dialogue system of SHFR-Ⅲ humanoid robot is designed based on the SEQ2 SEQ network model.The sentiment analysis model for robot interaction is constructed in the form of ensemble learning.In the design process of sentiment analysis model,the performance of multiple methods is tested.Through the integration of multiple models,the accuracy of text-based sentiment analysis is improved.In the dialogue system design,the maximum mutual information is used as the objective function of the neural network model.更多还原