【作者】 黄琴；

【导师】 杨贯中；

【作者基本信息】 湖南大学 ， 计算机应用技术， 2004， 硕士

【摘要】 论文针对网络学习个性化服务的需求，并通过对Web数据挖掘技术的深入研究，构建了个性化网络学习的模型以及学习者个性化信息模型，进而提出了网络学习个性化推荐的思想，并设计了其原型系统——PRELS。该原型系统主要将Web数据挖掘技术应用于网络学习环境中对学习者进行学习资源、学习行为或者捷径的推荐，使得学习者进行网络学习的绩效得到提高，进而改善了网络学习的质量，也满足了学习者个性化的需求。 论文运用将Web内容挖掘，Web结构挖掘，以及Web使用挖掘相结合的方法实现网络学习个性化推荐。该网络学习个性化推荐系统(PRELS)运用Web使用挖掘技术，通过分析学习者会话的特征以及学习者会话之间的相似性，向学习者提供个性化服务。其中，论文提出了关联数据模糊竞争聚类(fuzzy competitive clustering for relation data，FCCRD)算法，该FCCRD聚类算法是适应于网络学习环境特征，并针对没有明确特征的学习者行为数据而提出的，不仅如此，该算法还将不是严格的欧几里得的学习者会话间相似度数据转换为欧几里得数据，进而提高了聚类的效果。在运用Web使用挖掘技术的基础上，该推荐系统还进一步结合了Web内容挖掘技术和Web结构挖掘技术，提出运用上述技术实现个性化推荐的算法，并将其与Web使用挖掘技术相结合，共同组成了网络学习个性化推荐算法，其目的在于运用内容特征聚类得出学习者所访问的Web页面内容之间的相关性，以及运用关联规则分析得出网络学习Web站点的结构特征，进一步提高个性化推荐的精确度和全面性。 最后，论文通过标记语言XML及程序设计语言JAVA实现了上述网络学习个性化推荐系统(PRELS)。总之，论文提出的网络学习个性化推荐系统(PRELS)克服了需要用户提供主观的评价信息，不能处理大规模的数据量，用户的评价信息可能会过时，使用时不方便等缺陷，从而体现了该网络学习个性化推荐系统的优越性。更多还原

【Abstract】 The thesis designs the model of personality e-learning and the model of students’ information, and raises the idea of personality recommendation system of e-learning ( PRELS ) by studying the requirement of personality service of e-leaming. The system applies Web data mining technology in the recommendation of resource and behavior to the students. It makes the students increase the performance of e-learning, makes the students improve the quality of e-learning, and makes the students satisfy their requirement.Foremost, the PRELS finds out the usage pattern of the students, and provides personality service to the students, applying the Web data mining technology with the association rule and clustering technology. The system combines Web usage mining with Web content mining and Web structure mining, and increases the coverage and accuracy of personality recommendation using the relativity of the content of Web pages. Moreover, the system combines Web data mining technology with the FCCRD algorithm and the association rule to form the personality recommendation algorithm of e-learning. The FCCRD algorithm aims at the data of the students’ behaviors suiting for the character of e-learning. The FCCRD algorithm transforms the data of comparability of the students’ sessions to the Euclid data to improve the effect of clustering.Finally, the PRELS realizes by XML and JAVA. The thesis studies the evaluation method of the superiority of personality recommendation system and the performance test method of recommendation algorithms. And the PRELS overcomes the limitations of providing information of subjective evaluation by user, the limitations of not dealing with large numbers of data, the limitations of the outdated information of evaluation, and the limitations of using inconveniently. Ultimately, the personality recommendation algorithm incarnates the superiority of the PRELS.更多还原