【作者】 刘洋；

【导师】 苑津莎；

【作者基本信息】 华北电力大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 以机器人为代表的人工智能是未来社会发展的方向,而知识工程在人工智能领域占据着基础地位,其重要程度不言而喻。知识工程与自然语言处理领域吸引了大批的科研和学者投身其中。文献作为自然语言文字最重要的组织形式,是获取知识和学习经验的主要来源之一。机器人若能够自动化地获取文献中的经验、存储经验,在人们遇到同类事件时,能帮助人们采取高效的应对方法。机器人是以计算机为大脑的,机器人的大脑不应仅会“计算”,还应该懂得语言、能够阅读文献和学习文献中有用的经验。本文研究机器人阅读文献提取经验的方法,首先从基础的知识表示出发,再到逐渐深入探讨采用何种方法和如何设计模型来存储机器人获取到的经验。主要内容如下:(1)总结有关知识表示的具体方法,详细介绍了与论文相关的知识表示和存储方式。(2)结合自然语言处理中的分词技术和语法分类模型研究中文语句存储的问题。(3)建立了机器人阅读文献所需的事件层和经验层存储方案。构造了事件存储模型。采用将句子划分语义块并按照语义块成分组合成事件的方法,进行事件存储。(4)给出了事件知识转化为经验知识的步骤,设计了经验存储模型的树形结构。为使机器人有效查找经验,提出了一种利用相似度计算的方法查找与待解决事件最相似经验的方法。更多还原

【Abstract】 Artificial intelligence represented by robots is the direction of future social development,and knowledge engineering occupies a basic position in the field of artificial intelligence,and its importance is self-evident.The fields of knowledge engineering and natural language processing have attracted a large number of scientific researches and scholars to join them.As the most important organizational form of natural language and writing,literature is one of the main sources of knowledge and learning experience.If robots can automatically acquire and store experience in the literature,they can help people to adopt efficient response methods when they encounter similar incidents.A robot uses a computer as its brain.The robot’s brain should not only "calculate",but should also understand language,be able to read literature and learn useful experiences in literature.This paper studies the method of extracting experience from the robot reading literature,starting from the basic knowledge representation,and then gradually discussing which method and how to design the model to store the experience obtained by the robot.The main contents are as follows:(1)Summarize the specific methods of knowledge representation,and introduce in detail the knowledge representation and storage methods related to the paper.(2)Combining word segmentation technology in natural language processing and grammatical classification model to study the storage of Chinese sentences.(3)The event layer and experience layer storage scheme required for robot reading literature are established.The event storage model has been constructed.Event storage is carried out by dividing sentences into semantic blocks and combining them into events according to the components of semantic blocks.(4)The steps of transforming event knowledge into experience knowledge are given,and the tree structure of the experience storage model is designed.In order to enable the robot to find the experience effectively,a method using similarity calculation is proposed to find the experience that is most similar to the event to be solved.更多还原