导研系统的设计与开发

发布时间：2018-05-06 05:30

本文选题：导研系统 + 协同过滤算法　；参考：《华中师范大学》2017年硕士论文

【摘要】：随着教育信息化的普及,导师制研究生培养系统在研究生教育中崭露头角。导师制研究生培养系统以提供丰富的学习资源和便捷的交互方式成为研究生提高科研能力的重要途径之一。它不仅为研究生提供自主学习的平台,还帮助导师开展研究生科研交流工作。在大数据环境下,研究生往往会面临“信息沉溺”问题,使研究生难以在“信息海”中找到所需的学习资源。同时,学习资源的多样性难以得到保证。如何帮助用户在众多学习资源中快速找到所需的资源,并从多方面扩展相关研究点,即如何建立导研系统,成为本文探索和研究的主要内容。本文首先进行了加权Slope One协同过滤算法的相关研究,一方面,融入用户相似度算法,增强兴趣相似度高的用户对推荐结果的影响,另一方面,融入时间加权算法,降低较早的评分对预测评分的权重,从而根据用户兴趣偏好预测用户对目标资源的评分并得到推荐列表,主动向用户推送学习资源。其次利用信息聚合技术,在六个维度的基础上建立基于主题的学习资源聚合模型,扩展研究生学习广度,为后续科研工作做铺垫。其中引入网络爬虫技术,完成学习资源的自动入库,减轻人工上传学习资源的工作量。同时,采用协同工作工具Tower提高科研团队的协作效率,加强导师对研究生科研任务的管理。基于以上理论研究,本文采用MVC设计模式,运用Java语言、MySQL数据库和Tomcat服务器等,设计并实现了辅助研究生培养的导研系统,其功能包括资源、实战、计划、猿问和周记五大模块。MovieLens数据集上的Python离线实验结果说明,相对于传统的加权Slope One算法,本文的改进算法在一定程度上提高了推荐资源的质量。
[Abstract]:With the popularization of educational informatization, the tutorial graduate training system is emerging in graduate education. Tutorial system has become one of the important ways for graduate students to improve their scientific research ability by providing abundant learning resources and convenient interaction. It not only provides a platform for graduate students to learn independently, but also helps tutors to carry out graduate research exchanges. In the environment of big data, graduate students often face the problem of "information indulgence", which makes it difficult for graduate students to find the necessary learning resources in "information sea". At the same time, the diversity of learning resources is difficult to guarantee. How to help users find the resources they need quickly, and how to expand the related research points from many aspects, that is, how to establish the research guidance system, has become the main content of this paper. This paper firstly studies the weighted Slope One collaborative filtering algorithm. On the one hand, it integrates the user similarity algorithm to enhance the influence of users with high interest similarity on the recommendation results, on the other hand, it integrates the time-weighted algorithm. By reducing the weight of the earlier score to the prediction score, the user's score on the target resource is predicted according to the user's interest preference and the recommendation list is obtained, and the learning resources are pushed to the user actively. Secondly, the information aggregation technology is used to establish a theme-based learning resource aggregation model on the basis of six dimensions to expand graduate students' learning span and pave the way for further scientific research. The technology of web crawler is introduced to complete the automatic storage of learning resources and reduce the workload of manual uploading of learning resources. At the same time, the cooperative work tool Tower is used to improve the cooperation efficiency of scientific research team and to strengthen the management of graduate research task by tutors. Based on the above theoretical research, this paper adopts the MVC design pattern, uses Java language to design and implement the guiding and research system for postgraduate training, whose functions include resources, actual combat, planning, etc. The results of Python off-line experiments on the five modules. Movie Lens dataset show that compared with the traditional weighted Slope One algorithm, the improved algorithm in this paper improves the quality of the recommended resources to a certain extent.
【学位授予单位】：华中师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.52

【参考文献】