Project Engage: Training Secondary Teachers to Deliver Computer Science and Engineering Instruction

培养能够支持高中生在严谨的学术计算机科学课程中取得成功的计算机科学教师是国家的需要。项目参与：培训中学教师提供计算机科学和工程教学将支持在德克萨斯州45所城市、郊区和农村学校（包括公立、私立、磁铁和特许学校）实施一门名为“在我们的数字世界中茁壮成长”（TODW）的新兴计算机科学原理课程。TODW课程将采用独特的双注册模式（高中和大学学分同步）。该项目将测试两种创新的专业发展技术，以扩大到更多的学校：翻转课堂技术，教师接受录制的视频专业发展，然后利用面对面的时间进行更多的实践活动；以及一个自动化系统，该系统将使用人工智能技术来支持高中教师和大学教师之间就学生作业的共同成绩进行谈判。该项目的研究议程将解决以下核心挑战：在计算机科学等专业知识有限的领域，我们如何最有效地扩展专业发展，使尽可能多的学校教师能够提供高质量的教学？STEM- c（科学、技术、工程和数学，包括计算机）伙伴关系项目支持STEM专家和K-12学校系统之间的研究驱动型伙伴关系，以实现更好的K-12级别STEM教育的制度变革。STEM-C伙伴关系的计算机科学教育扩展项目建立在UTeachEngineering：通过国家科学基金会的数学和科学伙伴关系计划培训中学教师以提供基于设计的工程教学伙伴关系的基础上。这个项目将产生为TODW提供高度可伸缩课程所需的科学基础和具体工件。工件将包括一个不同的课程，可伸缩的专业发展，和可伸缩的评估工具和过程。差异化课程将包括形成性评估，教师可以使用这些评估来为个别学生或整个班级的课程构建新的途径。评估将通过双盲协作审查（DBCR）评分工具进行，该工具将包括机器学习算法，帮助检测个体（新手）计算机科学教师和（专家）大学水平计算机科学教授的分数之间的一致差异。DBCR工具将使用机器学习技术，如基本文本特征（词袋），句法和语义建模（n-grams和Latent Dirichlet Allocation），以及针对现有工件语料库的聚类分析。由于学生工件的评估明确地与标准联系在一起，人类评估者也可以使用DBCR工具来指定支持行项目标准分数的特定特性。可用的功能将因项目而异，但将利用DBCR用户界面中可用的各种工具：(1)可选择的分类分类，(2)突出显示相关文本功能，以及(3)标记文档结构的注释，例如交叉引用和支持证据。项目评估将关注TODW课程的实施保真度和学生在实施TODW的课堂上的成果。如果成功，该专业发展可以作为许多领域教师专业发展的可扩展模式。

Developing computer science teachers who can support high school students in being successful in rigorous, academic computer science courses is a national need. Project Engage: Training Secondary Teachers to Deliver Computer Science and Engineering Instruction will support the implementation of an emerging Computer Science Principles course called Thriving in Our Digital World (TODW) in forty-five urban, suburban, and rural schools, including public, private, magnet, and charter schools, in Texas. The TODW course will be offered in a unique dual enrollment mode (concurrent high school and college credit). This project will test two innovative professional development techniques for expanding the reach to more schools: flipped classroom techniques, in which teachers receive recorded video-based professional development, and then use face-to-face time for more hands-on activities; and an automated system which will use artificial intelligence technology to support negotiation of common grades for student work between high school teachers and college faculty. This project's research agenda will address the following core challenge: In a field such as computer science where there is limited expertise, how can we most efficiently scale professional development so as many schoolteachers as possible can provide high quality instruction? The STEM-C (Science, Technology, Engineering, and Mathematics, including Computing) Partnerships program supports research-driven partnerships between STEM experts and K-12 school systems to bring about institutional change for better STEM education at the K-12 level. This STEM-C Partnerships' Computer Science Education Expansion project builds on prior funding of the UTeachEngineering: Training Secondary Teachers to Deliver Design-Based Engineering Instruction Partnerships through the National Science Foundation's Math and Science Partnership program. This project will produce the scientific foundation and the concrete artifacts needed to deliver a highly scalable curriculum for TODW. The artifacts will include a differentiated curriculum, scalable professional development, and scalable assessment tools and processes. The differentiated curriculum will include formative assessments which can be used by teachers to architect novel pathways through the curriculum for individual students or whole classes. Assessments will be created that can be delivered through a double-blind collaborative review (DBCR) rubric scoring tool, which will include machine learning algorithms that help detect consistent discrepancies between scores of individual (novice) computer science teachers and (expert) college-level computer science professors. The DBCR tool will use machine learning techniques such as basic text features (Bag of Words), syntactic and semantic modeling (n-grams and Latent Dirichlet Allocation), and cluster analysis against existing artifact corpora. Since the assessments of student artifacts are explicitly tied to rubrics, human evaluators may also use the DBCR tool to specify specific features that support line item rubric scores. The available features will vary from project to project but will leverage a variety of tools available in the DBCR user interface: (1) selectable categorical classifications, (2) the highlighting of relevant text features, and (3) annotations to mark document structures, such as cross references and supporting evidence. Project evaluation will focus on both implementation fidelity of the TODW course and student-level outcomes in classes that implement TODW. If successful, the professional development could serve as a model for more scalable teacher professional development in many domains.