【作者】 王勇；

【导师】 符红光；

【作者基本信息】 电子科技大学 ， 计算机软件与理论， 2015， 硕士

【摘要】 随着信息技术的发展,智慧教育和人工智能技术成为广泛关注的热点。传统教育受限于对于教师的依赖,学生离开老师的辅导之后无法提高学习效率,虽然现在的在线辅导系统能在一定程度上辅助学生完成自我测评任务,但是系统不能对学生在应用题中的答题给出个性化的辅导,导致了学生对学习中出现的问题不能及时的发现和纠正。因此,本文通过对在线辅导系统进行研究,解决教育领域的知识表示问题,同时能够对于文本结构的知识形式进行自动的获取和处理,并以此为基于构建一个适用于初等数学应用题的自动评测系统,为人工智能在智慧教育中的应用提供了一个新的思路。主要的研究内容如下:1、初等数学知识表示。本文基于初等数学题库中知识的呈现形式,提出了一种一阶逻辑的知识表示方法,该方法结合面向对象技术,能够表示平面几何和初等代数知识,约束规则库与概念知识的分离,使得知识库的扩展和知识推理更加方便,同时有效的解决传统一阶谓词逻辑表示方法中对于高阶知识无法表示的问题。2、文本知识分类和获取。本文提出一个以谓词为基础的文本知识获取的流程。该流程中主要包括对文本数据的预处理,其中包括中文分词、句子边界确定和异常数据处理,构建初等数学句模库,并使用句模库对句子的语义进行归处理,如何提取文本知识中的特征,怎样构建基于条件概率的决策树算法,并且用实验证明算法的有效性。整个算法流程能够对文本形式的自然语言进行处理,并且将文本数据信息提取为本文定义的知识表示形式。3、根据本文提出的知识表示方法和文本知识处理算法,构建初等数学自动评测系统。该系统能够对学生做题数据进行挖掘分析,能够获取解题过程中的知识信息,并与答案中的标准知识列表进行匹配,以此分析学生在解题过程中错误类型,系统将解答中的错误展示学生,帮助学生提高学习效率。更多还原

【Abstract】 With the development of information technology, Smart Education and AIT(artificial intelligence technology) becomes the hot spot of worldwide attention. Traditional education is limited for the dependency on teachers,which leads to students’ failure on self-study if without help from teachers. Although to a degree,some online studying systems help students for self-testing,the problems through the progress can’t be discovered and corrected on time. So,this thesis focus on the study of online testing systems. For the of educational knowledge,adaptive acquiring and conducting of information from text,an adaptive valuing system is established to apply to elementary mathematics,providing a new path for the application of Smart Education in AIT. The main content is described as follows:1.Knowledge presentation of elementary mathematics. Based on elementary mathematics,this thesis propose a presentation method of first-order logic. The method combines OOT(Object-Oriented Technology)with Plane geometry and elementary algebra knowledge,separating restrictive rules from concepts. This makes it convenient for the extension and reasoning of knowledge base. Also,it solves the previous problem of First-order Predicate Logic method unable to present HOS.2.Classification and access of text knowledge. This thesis proposes an acquisition process of text knowledge based on predicate. It mainly includes the preprocessing of data text in the process,which contained splitting Chinese word, defining sentence boundaries and erasing error data, constructing model library of elementary mathematics, and using the library for the sentence semantic processing. It also tells how to extract the features in the text knowledge, how to construct the decision tree algorithm based on conditional probability, and prove the validity of the algorithm by experiments.The whole algorithm processes the natural language of text,while extracting information of text data for the definition of knowledge representation which is defined in this paper.3.According to the knowledge representation and text processing algorithms proposed, adaptive testing system of elementary mathematics is build. The system can do problem analysis and data mining of students to acquire knowledge in the process of problem solving, which is matched with the standard answer, to analyze the error types during problem solving. Problems and errors will be presented to students, helping them to improve their study efficiency.更多还原