在汉语教学中，句式作为特定的成分搭配组合和高频的表达规则，已被广泛应用于教师授课和教材教案编写。而目前句式的提取和审定往往依赖于专家的主观判断，效率较低且难以保证提取结果的一致性和完备性。本文针对人工提取句式繁琐低效的问题，对融合职业中文领域语料特点的句式自动提取算法展开研究，提升了句式提取的效率和科学性，并将所提方法应用于教学实践，设计句式教学辅助系统为师生提供对照展示、例句检索和句式推荐等服务，具有理论意义和实践价值。主要内容和创新如下：
(1) 构建句式提取的基本工作框架。基于句本位语法理论，从句法教学和信息处理的实际需求出发，对句式的构成元素及其约束组合规律进行研究，定义句式的形式化组成规则。将句式提取任务分为抽取扩展形式表达式和句式筛选两个子任务，明确各步骤的任务目标、中间产物和工程实现流程。
(2) 提出了形式表达式的抽取与扩展算法。首先，提出了融合句本位分层析句思想的单层结构表达式生成方法。其次，采用复合策略对结构表达式进行扩展拆分。最后，应用后处理算法获取形式表达式，将其作为句式提取的重要中间产物加入候选句式集合，以供下一步筛选。
(3) 提出了基于最优阈值的序贯三支决策句式筛选算法。首先，选取形式表达式和结构表达式在不同维度下的属性并映射至决策信息表。其次，将最优阈值用于单步决策判断过程，提出了一种基于最优阈值的序贯三支决策句式筛选算法。最后，在六个专业的职业中文教材上进行对比实验，选取最优距离算子并验证了所提句式筛选算法的有效性。
(4) 研究提取句式的教学应用并设计教学辅助系统。研究结构框架分类、相似句式归并展示等教学应用算法，向师生清晰简明地展示提取获得的句式。设计句式教学辅助系统为用户提供原文例句检索、句式对照展示编辑和相似句式推荐等服务，为学生学习巩固和教师教案编写提供辅助。
综上所述，本文结合句本位分层析句、多粒度粗糙集、三支粒计算等理论和方法，面向职业中文教学需求，对句式自动提取算法及其应用进行研究，提升了句式提取的效率和科学性，丰富了句式的应用场景，为融合职业中文领域语料特点的句式自动提取提供了一种新的思路和方法。

In Chinese language teaching, sentence patterns have been widely used in teacher lectures and textbook lesson planning as a specific combination of components and high-frequency expression rules. At present, the extraction and validation of sentence patterns often rely on experts' subjective judgment, making it inefficient and difficult to ensure the consistency and completeness of the extraction results. In response to the tedious and inefficient manual extraction problem, this thesis uses an automatic extraction algorithm that integrates the characteristics of vocational Chinese language materials and applies the proposed method to teaching practice, improving the efficiency and scientificity of sentence pattern extraction. A teaching assistance system has been designed to provide teachers and students with services such as comparative display, example sentence retrieval, and sentence recommendation, which have theoretical significance and practical value. The main content and innovations are as follows:
(1) A basic framework for sentence pattern extraction. Based on the theory of Sentence-based Grammar and hierarchical sentence structure, starting from the practical needs of syntax teaching and information processing, the constituent elements and constraint combination rules of sentence structures have been studied. The formal composition rules of sentence structures are defined. The task of Sentence pattern extraction is divided into two subtasks: extracting and expanding formal expressions as well as sentence pattern selection. The task objectives, intermediate products, and engineering implementation process for each step are clarified.
(2) A formal expression extraction and expansion algorithm. First, a single-layer structural expression extraction method is proposed by integrating Sentence-based hierarchical analysis. Secondly, combined with the composite strategy, the expanding algorithm splits and derives the structural expression. Finally, the formal expressions, as intermediate products in sentence pattern extraction, are added to the candidate set by a post-processing algorithm for sentence pattern selection.
(3)    A sequential three-way decision algorithm based on an optimal threshold for sentence pattern selection. First, the attributes of the formal and structural expressions in different dimensions are selected and mapped to the decision information table. Secondly, the optimal threshold is applied to the one-step decision-making process, and a sequential three-way decision algorithm based on an optimal threshold for sentence pattern selection is proposed. Finally, a comparative experiment is carried out on vocational Chinese textbooks of six majors to select the optimal distance operator and verify the effectiveness of the proposed algorithm.
(4)    Research on the teaching application of extracted sentence patterns and the designing of a teaching assistant system. Teaching application algorithms such as structural framework classification and similar sentence structure merging are proposed to present the extracted sentence patterns clearly and concisely to teachers and students. The teaching assistant system is designed to provide users with services such as original sample sentence retrieval, sentence pattern comparison editing, and similar sentence pattern recommendation to assist students' learning consolidation and teachers' teaching plan compilation.
To sum up, combined with the theories and methods of Sentence-based hierarchical analysis, multi-granularity rough sets, and three-way granular computing, this thesis studies the automatic sentence pattern extraction algorithm and its application for the needs of vocational Chinese teaching. The efficiency and scientificity of sentence pattern extraction have been improved, and the application scenarios of sentence pattern have been enriched, providing a new idea and method for sentence pattern automatic extraction integrating the corpus characteristics in the field of vocational Chinese.