基于多源家校沟通数据的个体特征分析支撑平台

发布时间：2018-12-17 15:59

【摘要】：21世纪初,家校之间出现了全新的家校沟通方式,如移动校讯通、联通家校通和电信翼校通等。随着互联网的迅猛发展,家校沟通平台层出不穷,平台数据之间存在系统与语义层面的异构性,没有统一的标准,与其它家校沟通平台之间不兼容,势必会造成新的"数据孤岛"。目前传统数据仓库中基于ETL的异构数据源集成过程非常繁琐、语义异构难以处理等,并不能解决数据的共享、重用以及语义上的集成问题。另外本体能够准确表述概念语义且拥有强大的推理能力,基于本体生成规则来指导本体映射,避免了传统异构数据源集成方式的缺点。综上,本文提出了数据仓库中基于本体的异构数据源集成方案。该集成方案首先构建共享词汇库,采用自动标记数据存储算法标记数据源模式,生成应用本体,而后基于本体映射和推理,发现隐藏层次结构,并根据这些生成规则实现自动化ETL操作。而后将该集成方案应用于多源家校沟通数据的实际场景。案例表明,通过本体指导的ETL过程,解决了传统过程的不足,提高了数据源集成整合的效率。通过分析面向领域需求的插件架构,本文提出了基于插件技术的数据分析支撑平台,并按照功能模块化原则设计与实现。针对多数据源引入了多数据源适配框架,针对数据分析算法库提出了动态数据挖掘算法插件适配框架,并对算法插件实行动态加载与管理。最后以本文提出的基于插件技术的数据分析支撑平台为依托,通过基于本体的异构数据源集成之后的家校沟通数据目标数仓库为输入源,以主体学生、家长、教师的基本数据及行为数据进行个体特征进行分析与挖掘,针对结果进行科学管理与决策,那么势必能会对目前的教育格局带来不可估量的价值。
[Abstract]:At the beginning of 21 ~ (st) century, there are new ways of communication between home and school, such as mobile communication, Unicom and telecommunication. With the rapid development of the Internet, homeschooling communication platforms emerge in endlessly. There are system and semantic heterogeneity between platform data, there is no unified standard, and they are incompatible with other communication platforms. It is bound to create a new "data island". At present, the integration process of heterogeneous data sources based on ETL in traditional data warehouse is very complicated, semantic heterogeneity is difficult to deal with, and can not solve the problems of data sharing, reuse and semantic integration. In addition, ontology can accurately express conceptual semantics and has powerful reasoning ability. Ontology mapping is guided by ontology generation rules, which avoids the shortcomings of traditional heterogeneous data source integration. In summary, this paper proposes an ontology-based heterogeneous data source integration scheme in data warehouse. The integration scheme firstly constructs a shared vocabulary database, uses automatic tagging data storage algorithm to mark the data source pattern, generates application ontology, and then finds hidden hierarchy based on ontology mapping and reasoning. According to these generation rules, automatic ETL operation is realized. Then, the integration scheme is applied to the practical scenario of multi-source homeschooling data communication. The case shows that the ETL process guided by ontology solves the shortcomings of the traditional process and improves the efficiency of data source integration and integration. By analyzing the domain-oriented plug-in architecture, this paper proposes a data analysis support platform based on plug-in technology, and designs and implements it according to the principle of functional modularization. This paper introduces a multi-data source adaptation framework for multi-data sources, and proposes a dynamic data mining algorithm plug-in adaptation framework for the data analysis algorithm library, and implements dynamic loading and management of the algorithm plug-in. Finally, based on the data analysis support platform based on plug-in technology proposed in this paper, through the integration of heterogeneous data sources based on ontology, the data target number warehouse is used as the input source, and the main students, parents, as the input source. The basic data and behavior data of teachers are analyzed and excavated, and scientific management and decision are made according to the results, which is bound to bring inestimable value to the present educational pattern.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13

【相似文献】