足球视频搜索引擎中的用户偏好挖掘
发布时间：2018-12-18 13:45
【摘要】：目的互联网信息量的急速增长使得人们需要花费大量时间从搜索引擎召回的结果中浏览自身感兴趣的内容,结合用户的搜索日志信息和社交平台信息,提出一种分层的实时偏好挖掘模型,为用户提供个性化搜索服务。方法在系统分析偏好挖掘的国内外研究现状的基础上,针对足球视频,提出一种分层权重无向图(HWUG)用户偏好模型,充分考虑用户偏好之间的关联信息,通过获取用户在足球领域的显式和隐式反馈信息,提取反馈信息中的偏好标签和偏好动作,并引入时间衰减因子,实现用户足球偏好的实时计算。结果算法已经应用在搜球网(www.findball.net)的个性化检索结果排序和视频推荐上,并已经取得了很好的效果。结论实验结果表明,结合特定领域的知识,基于分层无向权重图模型的偏好挖掘算法能更准确和实时反映用户的足球偏好。
[Abstract]:Objective with the rapid growth of Internet information, people need to spend a lot of time browsing the content of their own interest from the results of search engine recall, combining the search log information and social platform information of users. A hierarchical real-time preference mining model is proposed to provide personalized search services for users. Methods on the basis of systematic analysis of the current situation of preference mining at home and abroad, a hierarchical weighted undirected graph (HWUG) user preference model is proposed for football video, which fully considers the related information between user preferences. By obtaining explicit and implicit feedback information from users in football domain, the preference tags and preference actions are extracted from the feedback information, and time decay factor is introduced to realize real-time calculation of users' soccer preferences. Results the algorithm has been applied to the sorting of personalized retrieval results and video recommendation of www.findball.net, and has achieved good results. Conclusion the experimental results show that the preference mining algorithm based on hierarchical undirected weight graph model can reflect users' soccer preferences more accurately and in real time.
【作者单位】： 华中科技大学计算机科学与技术学院;华中科技大学网络与计算中心;
【基金】：国家自然科学基金项目(61173114,61202300) 湖北省杰出青年基金项目(2010CDA084) 广东省产学研项目(2011B090400251) 中央高校基本科研业务费专项资金项目(2011QN057,2011TS094)
【分类号】：TP391.3
[Abstract]:Objective with the rapid growth of Internet information, people need to spend a lot of time browsing the content of their own interest from the results of search engine recall, combining the search log information and social platform information of users. A hierarchical real-time preference mining model is proposed to provide personalized search services for users. Methods on the basis of systematic analysis of the current situation of preference mining at home and abroad, a hierarchical weighted undirected graph (HWUG) user preference model is proposed for football video, which fully considers the related information between user preferences. By obtaining explicit and implicit feedback information from users in football domain, the preference tags and preference actions are extracted from the feedback information, and time decay factor is introduced to realize real-time calculation of users' soccer preferences. Results the algorithm has been applied to the sorting of personalized retrieval results and video recommendation of www.findball.net, and has achieved good results. Conclusion the experimental results show that the preference mining algorithm based on hierarchical undirected weight graph model can reflect users' soccer preferences more accurately and in real time.
【作者单位】： 华中科技大学计算机科学与技术学院;华中科技大学网络与计算中心;
【基金】：国家自然科学基金项目(61173114,61202300) 湖北省杰出青年基金项目(2010CDA084) 广东省产学研项目(2011B090400251) 中央高校基本科研业务费专项资金项目(2011QN057,2011TS094)
【分类号】：TP391.3
【共引文献】
相关期刊论文 前10条
1 纪良浩;;协作过滤信息推荐技术研究[J];重庆邮电大学学报(自然科学版);2012年01期
2 王建雄;;图书馆知识管理协同环境的构建[J];沧桑;2008年04期
3 黄永锋;覃罗春;;一种有效缓解协同过滤推荐评价数据稀疏问题的算法[J];东华大学学报(自然科学版);2013年01期
4 王道平;李志隆;杨岑;;基于热门物品惩罚和用户兴趣变化的知识推送算法[J];系统工程;2014年01期
5 欧建斌;;个性化搜索引擎研究[J];微型机与应用;2010年11期
6 马丽;;基于组合加权评分的Item-based协同过滤算法[J];现代图书情报技术;2008年11期
7 姚忠;魏佳;吴跃;;基于高维稀疏数据聚类的协同过滤推荐算法[J];信息系统学报;2008年02期
8 蒋，
本文编号：2385937
本文链接：https://www.wllwen.com/kejilunwen/sousuoyinqinglunwen/2385937.html
