论文部分内容阅读
针对目前人机对话系统智能性较低、语句库领域受限的现状,设计了面向日常生活常用话题的机器人回复语句库,用于日常生活聊天.语句库以汉语句子类型中的句类为模型,构建了涵盖日常生活常用的28个话题范围的900句语句库.通过关键词语的替换可以扩展出更多的语句,从而用于更广泛领域的人机对话交流,增加了语句库的完善性,尽量避免了出现答非所问的情形.同时该构建方法降低了人工构建语句库的工作量,而且由于语句库较小,确保了系统实现的实时性.人机交互对比实验表明:该语句库实现了使用较少的语句就基本可以进行日常话题的对话,满意度高于对比网络聊天机器人小i,使得人机交互更加友好.
In order to solve the problem of the low intelligence and the limitation of sentence database, the robot reply sentence database for daily life is designed to chat in daily life.The sentence database is modeled by the sentences in Chinese sentence types , Building 900 sentence libraries covering 28 topics commonly used in daily life.Through the replacement of key words, more sentences can be extended, which can be used in the exchange of man-machine dialogue in a wide range of fields and the perfection of sentence bank , As far as possible to avoid answering questions.At the same time, the construction method reduces the workload of manual construction of sentence bank, and because the sentence bank is smaller, the real-time performance of the system is ensured.Comparison of human-computer interaction shows that the sentence bank realizes Use fewer sentences on the basic topics that can be carried out on the dialogue, satisfaction is higher than the Internet chat robot i, making human-computer interaction more friendly.