MOOC社区中基于移动情境感知的学伴推荐方法研究  

A learning partner recommendation method for MOOC community based on mobile context-awareness

作　　者：廖宏建[1] 曲哲[2] 胡莹[1] LIAO Hong-jian;QU Zhe;HU Ying(Network and Modern Educational Technology Center,Guangzhou 510006,China;Library,Guangzhou University,Guangzhou 510006,China)

机构地区：[1]广州大学网络与现代教育技术中心,广东广州510006 [2]广州大学图书馆,广东广州510006

出　　处：《广州大学学报（自然科学版）》2021年第2期77-85,共9页Journal of Guangzhou University:Natural Science Edition

基　　金：国家社科基金青年资助项目(15CTQ034)。

摘　　要：在大规模学员参与的MOOC社区中智能推荐适配的学习伙伴,对促进学习投入和提升课程完成率具有重要意义.基于探究社区理论模型分析MOOC社区知识地图结构,梳理影响学伴推荐的用户情境(学习投入、评分、时间及位置)和社会情境(交互行为、强度及时效),相应地构建用户情境信任度和社会情境信任度;在社会情境信任度构建中,引入信任奖励因子和时间衰减因子,反映MOOC社交信任的动态性;最后融合用户情境和社会情境计算动态综合信任度,通过信任传播构建MOOC社区复杂信任网络,基于此为学员生成Top-N推荐列表.实验表明,融合用户情境和社会情境的学伴推荐方法,能充分挖掘学员兴趣偏好及交互行为变化所反映的动态信任关系,在高动态性和移动性的MOOC社区中具有较好的推荐效果;同时,可以通过调整多维情境的权重值实现在不同推荐场景的灵活应用.Intelligent learning partner recommendation is of great significance in promoting learning engagement and improving course completion rate in a large-scale MOOC community.Based on Community of Inquiry(COI),the structure of knowledge map in MOOC community was analyzed.As user context and social context were important factors which influence learning partner recommendation,degree of trust based on user context and social context were constructed correspondingly.In terms of social context trust degree,trust reward factor were time attenuation factor were introduced to reflect the dynamic nature of MOOC social trust.Finally,the Top-N recommendation list was generated for learners through the construction of a complex trust network in the MOOC community based on trust propagation.According to the research results,learning partner recommendation method,which integrates user context and social context,can fully mine the dynamic trust relationship reflected by the changes in students interests and interaction behavior.This method s effect is greatly recommended particularly in a high dynamic and mobile MOOC community.The method can also be applied to different recommendation scenarios flexibly by adjusting the weight value of the multi-dimensional context.

关 键 词：MOOC社区 情境感知 信任网络 知识地图 学伴推荐系统 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]