智慧教育大数据整体建设方案

一、 设计思路教育大数据技术为教学、学习及管理决策等教育活动提供了全新的科学工具，并有力地推动了教育的变革。1 聚焦数据价值教育数据采集与深度分析技术全覆盖教育业务各应用系统中，涵盖教学、管理、科研、培训等，既注重相关关系的识别，又强调因果关系的确定，通过数据分析技术发现教育中实际存在的问题，辅助用户解读和理解数据分析结果，更准确评价当前现状，预测未来趋势。将数据分析的结果融入学校的日常管理与服务之中，为师生提供精细化与智能化服务。2 坚持融合创新发挥技术优势，变革传统模式，推进新技术与教育教学的深度融合，真正实现从融合应用阶段迈入创新发展阶段。全面收集、准确分析、合理利用教育大数据，从“基于有限个案”向“基于全面数据”转变，推动教育决策从经验型、粗放型向精细化、智能化转变。3 提升师生素养整合多元应用，提供丰富、多样、个性化专业服务，提升师生信息素养，推动从技术应用向能力素质拓展，不再仅仅注重学生的学习成绩，而更加关注身心健康、学业进步、个性技能、成长体验等方面。建立师生良好信息思维，培养应用信息技术解决教学、学习、生活中问题的能力。4 转化思维模式课堂作为推动学校内涵发展的主阵地，通过大数据应用对教师进行课堂观察、数据采集和分析，得出测评结果，然后制定相应的提升措施，不断促进教师教学水平的提升。推动大数据技术与教师教育产生深度融合，促进教师专业发展、教师教育教学全方位变革与创新发.二、建设目标1 打造教育大数据平台，统领教育应用教育大数据以“数据集中、信息共享、业务共通、应用统领”为支撑，建立教育数据应用平台；所有应用统一于大数据平台，实现数据从各应用平台采集，汇聚到数据中心统一存储，应用层统一调用，分场景进行加工处理的模式。教育大数据平台，为教育管理者、教师、家长、学生等不同对象提供多层次、全方位的综合应用服务，综合构建“教、学、管、评、测、练”与“教育管理机构、学校、老师、家长、学生”相结合的多维教育大数据信息化体系。全面打通用户的基础数据，融合学业数据、教育管理综合信息、教师信息等大数据，进行全面多维的各类分析、数据透视。教育管理者、教师、学生、家长都能通过教育大数据平台，根据角色、权限及应用场景的不同，享用平台中各类教育应用。2 提供个性化教学服务通过对学生历年学业成绩、课程选修、活动参与等数据分析，除了追踪学生学业进步情况外，还可以从中分析不同学生的学习需求和风格，进而提供适应学生特点的个性化教学。一是通过数据分析对学习困难学生进行干预，教师通过学生数据系统监控学生学业表现进行干预性指导。二是获得学生学习结果的即时智能反馈。通过课堂行为记录与分析工具，教师可以及时获得学生学习情况并调整教学活动。三是在学生选择辅修课程或课外项目时，大数据技术可以提供适合学生的个性化建议。四是基于大数据分析改进日常教学工作。教师可以通过分析学生社交行为数据，更有效地开展团队和小组学习，优化学习计划和日程安排。3 变革教与学发展模式，提升师生数据素养对教师教学日志数据、教学资源数据、教学互动数据、教学评价数据、教学效果数据、教师继续教育数据、教学工具使用数据等日常教学过程、行为、结果数据的深入分析与挖掘建立教师数据素养，帮助教师更好地获得学生反馈，发现每位学生的兴趣点和薄弱点，以优化教学模式，改进教学策略，实现个性化教学；有助于教师有效预测学生考试成绩及发展趋势，及时干预并指导学生的学习与发展；有助于教师对学生做出全面客观的评价，推动教育评价方式从“经验主义”走向“数据主义”；有助于教师的教育决策更加科学准确，提高工作效率与学生的学习成绩；有助于教师发现自身专业技能的不足和问题，提升专业能力和研究水平，适应数据驱动教学时代的新要求。对学生日志数据、成绩预警数据、师生评价数据、在线话语数据、伦理隐私数据、多模态数据等六类数据的进行深入的挖掘分析,建立学生数据素养。培养学生在数据感知和采集、组织和管理、处理与分析、共享与协同创新等方面的能力，以及在数据的生产、管理和发布过程中的道德与行为规范。帮助学生更好地获取学习反馈，发现自身学习的优劣势，优化学习方式方法，实现精细化学习；帮助学生更好的预估学习发展趋势，指导学生做好学业生涯规划；帮助学生更精确的进行学习过程跟踪与学习过程评价，为学生综合素质评价提供数据支撑；帮助学生发现和学习高效学习方式方法，提升自适应学习能力，培养终身学习习惯。4 促进学校教育信息化发展，破解发展难题为了推进大数据应用服务驱动地区教学的快速发展，教育行政部门、教育大数据服务企业、中小学校应当协同发力，重点从五个方面推进实施，包括：开展数据素养专题培训，提高教师数据意识与数据处理能力；打造基于大数据的智慧学习平台，支撑教师开展数据驱动的精准教学；开展数据驱动教学示范项目，探索数据驱动教学新模式；构建数据驱动教学实践共同体，传播数据驱动教学文化；开展数据驱动教学专题研究，引领数据驱动教学持续深入发展。5 挖掘大数据推动教育研究转型现代教育运用实证数据研究教育具体问题，再基于研究结果指导政策与实践。大数据技术为大规模教育实证研究提供便利，推动教育研究转型为“数据密集型科研”。一是利用纵向数据开展长期性追踪研究。维格多以北卡罗莱州1500名教师为对象，跟踪分析1997—1998学年至2007—2008学年学生学业测评结果对教师工资薪酬的影响，提出基于绩效考核的教师酬金改革建议。二是开展大规模横向比较研究。美国“全国学生中心”(NSC)开展的大学阶段学业成就与高中阶段学业间关系研究，依托各州和联邦教育数据库，以全美92%的在校大学生为对象，通过了解不同学校入学情况、学生人口特征与大学入学的关系、高中阶段学业成绩相似学生在大学后的学业表现等内容，分析高中阶段学习对大学学业的影响。这种跨年度的追踪研究和大规模的横向比较研究，没有大数据的支撑是难以实现的。6 基于大数据推动教育科学决策全面的地区基础教育质量监测不仅让决策者了解教育的整体状况和变化趋势，还通过分析家庭背景、教育项目和学校教学与学生学业成绩间的关系，进而影响地区教育资源分配与资助性项目的实施。一是运用教育大数据规划学校布局与资源分配，地区和学校通过分析学生人口学数据，得出本地区学龄人口变动趋势，从而科学规划本地区学校布局与资源分配。二是改进学校绩效评估办法。基于学校整体与学生个体学业数据，评价学校的办学质量或项目实施质量，分析学校的优势与弱势领域。三是推动家校合作。通过使用智慧课堂反馈工具，教师可以实时上传本节课学生课堂表现和任务完成情况，学校和家长借此可以及时了解并与校方交流学生情况。四是提高学校管理效率。在学生出勤、用餐及校车运营等活动中使用学生管理软件，自动记录并通过数据分析提出改进方案。五是改革教师评聘方式。通过分析教师任教学生的学业成绩以及教师的职业信仰、专业发展、社会服务等指标，科学评估教师专业水平与发展潜能。7 开展教育大数据服务，惠及全体教育参与者通过涵盖K12范围内的小学、初中、高中全线贯通的区域教育生态监控大数据平台，全面跟进诊断班级、学校、区域教育的教与学现状；为全区各级各类教育主管部门实时分析全区域、各校教学质量，并智能提供教育优化方案，智慧化跟进全区基础教育质量；为全区所有教师提供精准教学和改进依据、为全区学生和家长提供了解自己学习现状的途径及改进渠道、资源和方法；提供基于移动互联网环境下的教与学内容探讨、交流互动的互动平台，形成教师、学生、家长三个群体之间的兴趣聚合和广泛讨论。三、 总体框架以满足大数据平台、应用、服务建设要求为基础，整合现有软硬件资源，构建数据机房计算、存储、网络、容灾、安全等软硬件基础设施环境；建设涵盖核心基础数据库、数据采集、数据共享、数据挖掘、数据控制、数据服务的数据处理中心大数据管控平台以及数据标准、数据安全等管理体系；聚合已建设应用管理系统，打造新型大数据应用服务系统，构建教学服务体系。全地区、全学校、全学段、全过程采集教育各环节数据单元，整合教育元数据内容形成教育数据集，打通教育各应用系统及管理口之间数据壁垒，形成教育数据链条。分析、挖掘、预警教育决策信息，聚类、提取、发现教、学、管各类行为习惯数据，全面数据化指导教育业务管理；精准、智能化提升教育质量水平；智慧、个性化培养教与学行为习惯。四、技术路线 大数据技术教育管理业务系统的各类应用将产生大量的业务数据。数据来源非常丰富且数据类型多样，存储和分析挖掘的数据量庞大，对数据展现的要求较高，并且很看重数据处理的高效性和可用性。大数据技术根植于云计算，着眼于“数据”，关注实际业务，提供海量数据采集、分析、挖掘、展现等方面的支持，看重的是海量数据存储管理能力。对结构化和非结构化混合的大数据，采用MPP 并行数据库集群与Hadoop集群的混合来实现对百PB 量级、EB量级数据的存储和管理。用MPP 来管理计算高质量的结构化数据，提供强大的SQL和OLTP型服务，Hadoop实现对半结构化和非结构化数据的处理，以支持诸如内容检索、深度挖掘与综合分析等新型应用。数据融合技术数据融合技术是指利用计算机对按时序获得的若干观测信息，在一定准则下加以自动分析、综合，以完成所需的决策和评估任务而进行的信息处理技术。为共享数据中心、主数据库、数据集市、数据仓库，提供统一的数据标准规范和数据接口，进行统一的数据集成与接入，实现从数据标准到元数据管理、主数据管理、数据仓库的自动同步，对各种信息源给出的有用信息的采集、传输、综合、过滤、相关及合成，以便辅助人们进行态势/环境判定、规划、探测、验证、诊断，并为可视化数据分析提供数据支撑。 大数据分析与可视化大规模数据的可视化主要是基于并行算法设计的技术，合理利用有限的计算资源，高效地处理和分析特定数据集的特性。通常情况下，大规模数据可视化的技术会结合多分辨率表示等方法，以获得足够的互动性能。数据分析和可视化基于计算处理层。分析包括简单的查询分析、流分析以及更复杂的分析(如机器学习、图计算等)。查询分析多基于表结构和关系函数，流分析基于数据、事件流以及简单的统计分析，而复杂分析则基于更复杂的数据结构与方法，如图、矩阵、迭代计算和线性代数。一般意义的可视化是对分析结果的展示。但是通过交互式可视化，还可以探索性地提问，使分析获得新的线索，形成迭代的分析和可视化。 微服务架构微服务架构是一项在云中部署应用和服务的新技术，一个应用是由多个小的、相互独立的、微服务组成，这些服务运行在自己的进程中，开发和发布都没有依赖。不同服务通过一些轻量级交互机制来通信，例如 RPC、HTTP 等，服务可独立扩展伸缩，每个服务定义了明确的边界，不同的服务甚至可以采用不同的编程语言来实现，由独立的团队来维护。

1. Design Philosophy
Educational big data technology provides brand-new scientific tools for educational activities such as teaching, learning, and management decision-making, and has strongly promoted educational reform.
1.1 Focus on Data Value
Cover all application systems of educational services through data collection and in-depth analysis technologies, including teaching, management, research, training, etc. While focusing on the identification of correlations, it also emphasizes the determination of causal relationships. Through data analysis technologies, we can identify practical problems in education, assist users in interpreting and understanding data analysis results, more accurately evaluate the current situation, and predict future trends. Integrate the results of data analysis into the daily management and services of schools, and provide refined and intelligent services for teachers and students.
1.2 Adhere to Integration and Innovation
Leverage technological advantages to transform traditional models, promote the in-depth integration of new technologies with education and teaching, and truly move from the integration application stage to the innovative development stage. Collect comprehensively, analyze accurately, and rationally utilize educational big data, shift from "based on limited cases" to "based on comprehensive data", and promote the transformation of educational decision-making from empirical, extensive mode to refined and intelligent mode.
1.3 Improve Teachers' and Students' Literacy
Integrate diverse applications to provide rich, diverse, and personalized professional services, improve the information literacy of teachers and students, promote the expansion from technical application to ability and quality, no longer only focus on students' academic performance, but pay more attention to physical and mental health, academic progress, individual skills, growth experience, etc. Establish good information thinking among teachers and students, and cultivate the ability to apply information technology to solve problems in teaching, learning, and life.
1.4 Transform Thinking Mode
The classroom, as the main front for promoting the connotative development of schools, conducts classroom observation, data collection and analysis of teachers through big data applications, obtains evaluation results, then formulates corresponding improvement measures, and continuously promotes the improvement of teachers' teaching level. Promote the in-depth integration of big data technology and teacher education, and promote the professional development of teachers and all-round reforms and innovative development of teacher education and teaching.

2. Construction Goals
2.1 Build an Educational Big Data Platform to Unify Educational Applications
Educational big data is supported by "data centralization, information sharing, business interconnection, and application unification", and an educational data application platform is established; all applications are unified under the big data platform, realizing the mode of collecting data from various application platforms, aggregating them to the data center for unified storage, and the application layer calls them uniformly for processing and analysis in different scenarios. The educational big data platform provides multi-level and comprehensive application services for different objects such as educational managers, teachers, parents, and students, and comprehensively builds a multi-dimensional educational big data information system that combines "teaching, learning, management, evaluation, testing, practice" and "educational management institutions, schools, teachers, parents, students". Fully connect users' basic data, integrate big data such as academic data, comprehensive educational management information, and teacher information, and conduct comprehensive multi-dimensional analysis and data perspective. Educational managers, teachers, students, and parents can access various educational applications on the platform according to their roles, permissions, and application scenarios.
2.2 Provide Personalized Teaching Services
Through data analysis of students' annual academic performance, course selection, activity participation, etc., in addition to tracking students' academic progress, we can also analyze the learning needs and styles of different students, and then provide personalized teaching adapted to students' characteristics. First, conduct interventions for students with learning difficulties through data analysis: teachers can monitor students' academic performance through the student data system and provide intervention guidance. Second, obtain real-time intelligent feedback on student learning results. Through classroom behavior recording and analysis tools, teachers can timely obtain students' learning status and adjust teaching activities. Third, when students choose minor courses or extracurricular projects, big data technology can provide personalized recommendations suitable for students. Fourth, improve daily teaching work based on big data analysis. Teachers can analyze students' social behavior data to carry out team and group learning more effectively, optimize learning plans and schedules.
2.3 Reform Teaching and Learning Development Models and Improve Teachers' and Students' Data Literacy
Conduct in-depth analysis and mining of daily teaching process, behavior, and result data, including teachers' teaching log data, teaching resource data, teaching interaction data, teaching evaluation data, teaching effect data, teacher continuing education data, teaching tool usage data, etc., to establish teachers' data literacy, help teachers better obtain student feedback, identify each student's interests and weak points, optimize teaching models, improve teaching strategies, and realize personalized teaching; it helps teachers effectively predict students' exam scores and development trends, intervene in time and guide students' learning and development; it helps teachers make comprehensive and objective evaluations of students, promoting the transformation of educational evaluation methods from "empiricism" to "datism"; it helps teachers make more scientific and accurate educational decisions, improve work efficiency and students' academic performance; it helps teachers discover the deficiencies and problems in their own professional skills, improve professional abilities and research levels, and adapt to the new requirements of the data-driven teaching era.
Conduct in-depth mining and analysis of six types of data of students, including student log data, performance early warning data, teacher-student evaluation data, online discourse data, ethical and privacy data, and multimodal data, to establish students' data literacy. Cultivate students' abilities in data perception and collection, organization and management, processing and analysis, sharing and collaborative innovation, etc., as well as moral and behavioral norms in the process of data production, management and release. Help students better obtain learning feedback, identify their own learning advantages and disadvantages, optimize learning methods, and realize refined learning; help students better predict learning development trends and guide students to make academic career plans; help students more accurately track and evaluate the learning process, and provide data support for students' comprehensive quality evaluation; help students discover and learn efficient learning methods, improve adaptive learning ability, and cultivate lifelong learning habits.
2.4 Promote the Development of School Educational Informatization and Resolve Development Difficulties
In order to promote the rapid development of regional teaching driven by big data application services, educational administrative departments, educational big data service enterprises, and primary and secondary schools should work together, focusing on promoting implementation from five aspects: carry out special training on data literacy to improve teachers' data awareness and data processing capabilities; build a smart learning platform based on big data to support teachers in carrying out data-driven precise teaching; carry out data-driven teaching demonstration projects to explore new models of data-driven teaching; build data-driven teaching practice communities to disseminate data-driven teaching culture; carry out special research on data-driven teaching to lead the continuous in-depth development of data-driven teaching.
2.5 Leverage Big Data to Promote the Transformation of Educational Research
Modern education uses empirical data to study specific educational issues, and then guides policies and practices based on research results. Big data technology provides convenience for large-scale educational empirical research, promoting the transformation of educational research into "data-intensive research". First, use longitudinal data to carry out long-term tracking research. Vigdor took 1,500 teachers in North Carolina as subjects, tracked and analyzed the impact of students' academic evaluation results from the 1997-1998 academic year to the 2007-2008 academic year on teachers' salaries, and put forward reform recommendations for teacher remuneration based on performance appraisal. Second, carry out large-scale cross-sectional comparative research. The U.S. "National Student Clearinghouse" (NSC) conducted a study on the relationship between college academic achievement and high school academic achievement, relying on state and federal educational databases, targeting 92% of college students across the United States, analyzing the impact of high school learning on college academic performance by understanding the relationship between school enrollment status, student demographic characteristics and college enrollment, and the academic performance of students with similar high school academic performance in college. Such cross-year tracking research and large-scale cross-sectional comparative research cannot be realized without the support of big data.
2.6 Promote Scientific Educational Decision-Making Based on Big Data
Comprehensive regional basic education quality monitoring not only allows decision-makers to understand the overall status and changing trends of education, but also analyzes the relationship between family background, educational projects, school teaching and students' academic performance, thereby affecting regional educational resource allocation and the implementation of subsidized projects. First, use educational big data to plan school layout and resource allocation: regions and schools can analyze student demographic data to obtain the changing trend of school-age population in the region, so as to scientifically plan the regional school layout and resource allocation. Second, improve school performance evaluation methods: evaluate the school's running quality or project implementation quality based on the overall school and individual student academic data, and analyze the school's strengths and weak areas. Third, promote home-school cooperation: through the use of smart classroom feedback tools, teachers can upload students' classroom performance and task completion status of this class in real time, and schools and parents can timely understand and communicate with the school about students' situation. Fourth, improve school management efficiency: use student management software in student attendance, dining, school bus operation and other activities to automatically record and put forward improvement plans through data analysis. Fifth, reform teacher evaluation and employment methods: scientifically evaluate teachers' professional level and development potential by analyzing the academic performance of students taught by teachers and indicators such as teachers' professional beliefs, professional development, and social services.
2.7 Carry out Educational Big Data Services to Benefit All Educational Participants
Through the regional educational ecology monitoring big data platform covering all K12 stages (primary school, junior high school, senior high school), comprehensively track and diagnose the teaching and learning status of classes, schools, and regional education; provide real-time analysis of the teaching quality of the entire region and various schools in the region, and intelligently provide educational optimization plans, and intelligently follow up the quality of basic education in the region; provide precise teaching and improvement basis for all teachers in the region, provide ways and improvement channels, resources and methods for all students and parents in the region to understand their own learning status; provide an interactive platform for discussion, communication and interaction of teaching and learning content based on the mobile Internet environment, forming interest aggregation and extensive discussions among the three groups of teachers, students, and parents.

3. Overall Framework
Based on meeting the construction requirements of big data platforms, applications and services, integrate existing software and hardware resources to build a software and hardware infrastructure environment such as data computer room computing, storage, network, disaster recovery, and security; build a big data management and control platform for data processing centers covering core basic databases, data collection, data sharing, data mining, data control, and data services, as well as management systems such as data standards and data security; aggregate the built application management systems, build a new big data application service system, and construct a teaching service system.
Collect data units in all links of education across the entire region, all schools, all school stages, and the entire process, integrate educational metadata content to form educational data sets, break through data barriers between various educational application systems and management ports, and form an educational data chain. Analyze, mine, and early warn educational decision-making information, cluster, extract, and discover various behavioral habit data of teaching, learning, and management, and comprehensively guide educational business management through digitization; accurately and intelligently improve the quality of education; intelligently and personalized cultivate teaching and learning behavioral habits.

4. Technical Route
4.1 Big Data Technology
Various applications of the big data technology education management business system will generate a large amount of business data. The data sources are very rich and the data types are diverse, the volume of stored, analyzed and mined data is huge, the requirements for data presentation are high, and great importance is attached to the efficiency and availability of data processing. Big data technology is rooted in cloud computing, focuses on "data", pays attention to actual business, provides support for massive data collection, analysis, mining, presentation, etc., and attaches great importance to massive data storage and management capabilities. For mixed structured and unstructured big data, a hybrid of MPP parallel database cluster and Hadoop cluster is adopted to realize the storage and management of data of hundreds of PB and EB levels. Use MPP to manage and calculate high-quality structured data, providing powerful SQL and OLTP services, and use Hadoop to process semi-structured and unstructured data to support new applications such as content retrieval, deep mining and comprehensive analysis.
4.2 Data Fusion Technology
Data fusion technology refers to the information processing technology that uses computers to automatically analyze and synthesize several observation information obtained in time sequence under certain criteria, to complete the required decision-making and evaluation tasks. It provides unified data standard specifications and data interfaces for shared data centers, master databases, data marts, and data warehouses, carry out unified data integration and access, realize automatic synchronization from data standards to metadata management, master data management, and data warehouses, collect, transmit, synthesize, filter, correlate and synthesize useful information from various information sources, so as to assist people in situation/environment judgment, planning, detection, verification, diagnosis, and provide data support for visualized data analysis.
4.3 Big Data Analysis and Visualization
The visualization of large-scale data is mainly based on parallel algorithm design technologies, reasonably utilizing limited computing resources to efficiently process and analyze the characteristics of specific data sets. Generally, large-scale data visualization technologies combine methods such as multi-resolution representation to obtain sufficient interactive performance. Data analysis and visualization are based on the computing processing layer. Analysis includes simple query analysis, stream analysis, and more complex analysis (such as machine learning, graph computing, etc.). Query analysis is mostly based on table structures and relational functions, stream analysis is based on data, event streams and simple statistical analysis, and complex analysis is based on more complex data structures and methods, such as graphs, matrices, iterative computing and linear algebra. General visualization is the display of analysis results. However, through interactive visualization, you can also ask exploratory questions, so that the analysis obtains new clues, forming iterative analysis and visualization.
4.4 Microservices Architecture
Microservices Architecture is a new technology for deploying applications and services in the cloud. An application is composed of multiple small, mutually independent microservices, which run in their own processes without dependencies on development and release. Different services communicate through lightweight interaction mechanisms, such as RPC, HTTP, etc. Services can be independently expanded and scaled, each service defines clear boundaries, and different services can even be implemented in different programming languages and maintained by independent teams.

搜集汇总

背景与挑战

背景概述

该数据集详细介绍了智慧教育大数据平台的建设方案，旨在通过数据集中和共享，实现教育应用的智能化和个性化。方案涵盖了数据采集、分析、挖掘和可视化技术，以及如何通过这些技术提升教学质量和师生数据素养。