移动环境下的个性化学习资源推荐策略研究

发布时间：2018-11-26 15:22

【摘要】：近些年,移动智能终端大量普及,通信技术日新月异,互联网信息的主要传播媒介已经从传统的PC端转向移动端。虽然传播媒介发生了明显改变,但移动终端用户仍然面对着“信息爆炸”和“信息过载”这一互联网时代的顽疾,并且受限于移动设备计算能力低、显示面积小,使用环境复杂等特点,问题显得更加严重。个性化推荐技术提出和应用有效缓解了当前所面临的问题,其本质是根据用户的兴趣取向预测用户对资源的偏好,形成推荐列表。20多年的研究历史使得个性化推荐技术获得了较快发展,在电子商务等领域的成功应用使其研究取得了非凡成就。移动互联网的浪潮推动了移动领域的个性化推荐技术的研究和应用,但还没有达到很完善的地步,特别是具体领域的应用研究比较缺乏。本文围绕移动学习资源的个性化推荐展开移动环境下的推荐策略研究。文章针对性地介绍了移动学习资源推荐的国内外研究现状、移动学习的概念、情景感知服务体系以及目前存在的多种个性化推荐算法的实现原理和优缺点,这为展开研究活动提供了重要理论依据。文中重点论述了基于协同过滤技术的个性化推荐算法原理、算法缺陷及优化方法,并在此基础上提出了一种融合移动情景上下文的个性化移动学习资源推荐算法MCCF。该算法有效缓解了目前相关研究存在的两个问题:一是针对移动学习资源的个性化推荐中学习者移动情景上下文问题的研究,以提高推荐结果的准确率;二是针对移动终端计算性能的限制,必须考虑所采用的算法复杂度,需在推荐计算方面满足及时性的要求。本文提出的推荐策略将提取到的学习者移动上下文情景作为学习者的偏好特征,实现了U(学习者)×I(学习资源)×C(移动情景上下文)三维模型向U×I二维模型的转化,从而使得推荐结果能够应用传统的协同过滤算法得到。通过进一步研究发现,利用数据分析和数据挖掘技术能够得到学习者移动情景特征和资源特征的关联度,这种关联度能动态地反映用户偏好,从而优化最终推荐结果。MCCF算法既充分融入了学习者移动情景信息,同时,兼顾了算法的复杂度,最终的实验结果表明该算法在准确性和用户满意度上均优于传统算法,并且及时性较高,适合移动终端的推荐场景。
[Abstract]:In recent years, mobile intelligent terminals are widely used and communication technology is changing with each passing day. The main media of Internet information has changed from traditional PC terminal to mobile terminal. Although the media has changed markedly, mobile end users are still facing the "information explosion" and "information overload", which are the persistent diseases of the Internet era, and are limited by the low computing power of mobile devices and the small display area. The use of complex environment and other characteristics, the problem appears to be more serious. Personalized recommendation technology has effectively alleviated the current problems, and its essence is to predict the user's preference for resources according to the user's interest orientation. The research history of more than 20 years has made the personalized recommendation technology develop rapidly, and the successful application in the field of electronic commerce makes it make remarkable achievements. The wave of mobile Internet has promoted the research and application of personalized recommendation technology in mobile field, but it has not reached a very perfect stage, especially in specific fields. This paper focuses on the personalized recommendation of mobile learning resources in mobile environment. This paper introduces the current situation of mobile learning resources recommendation at home and abroad, the concept of mobile learning, the service system of situational perception, and the implementation principle, advantages and disadvantages of various personalized recommendation algorithms. This provides an important theoretical basis for carrying out research activities. In this paper, the principle, algorithm defect and optimization method of personalized recommendation algorithm based on collaborative filtering technology are discussed. On this basis, a personalized mobile learning resource recommendation algorithm, MCCF., which combines mobile context with mobile context, is proposed. This algorithm effectively alleviates two problems existing in the related research. One is the research on the context of mobile situation in the personalized recommendation of mobile learning resources to improve the accuracy of the recommendation results; Second, due to the limitation of mobile terminal computing performance, the complexity of the algorithm must be taken into account, and the requirement of timeliness should be satisfied in the aspect of recommended computing. The proposed recommendation strategy takes the extracted learner mobile context as the learner preference feature, and realizes the conversion of U (learner) 脳 I (learning resource) 脳 C (mobile context) 3D model to U 脳 I two-dimensional model. Thus, the recommended results can be obtained by using the traditional collaborative filtering algorithm. Through further research, it is found that using data analysis and data mining techniques can obtain the correlation degree of mobile scene features and resource features of learners, which can dynamically reflect user preferences. The MCCF algorithm not only fully integrates the learner mobile scene information, but also takes into account the complexity of the algorithm. The final experimental results show that the algorithm is superior to the traditional algorithm in accuracy and user satisfaction. And the timeliness is high, suits the mobile terminal recommendation scene.
【学位授予单位】：云南师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【相似文献】