A Unified Fusion Framework for Time-related Rank in Threaded Discussion Communities | |
You Q(游强)![]() ![]() ![]() | |
2014-11-24 | |
会议名称 | PAKDD 2014 International Workshops |
会议录名称 | Lecture Notes in Computer Science |
卷号 | Volume 8643 |
页码 | pp 513-524 |
会议日期 | 2014.5.14 - 2014.5.16 |
会议地点 | 台湾省台南县 |
出版者 | Springer International Publishing |
摘要 | We propose a unified fusion framework for time-related rank, applied to find valuable posts or recommend answers in threaded discussion communities. In our model, we simultaneously consider the special structure and semantics of threaded discussion communities. As for the structure, we construct a time-related rank model with respect to reply posts analysis and attain an initial rank result. Concurrently, we reconstruct semantic trees from raw statistical features (e.g. term frequency and document length) to latent semantics and topics. With a more robust similarity computation, we produce several semantic trees. For each tree, we again compute the time-related rank score and get a series of rank results. Finally, we fuse our results in the unified fusion framework incorporating quality measures to make a final decision. Our model can be easily extended when new features or models are added. Experimental results show that our model contributes satisfactory results. |
关键词 | Fusion Quality Measure Time-related Rank Information Retrieval |
DOI | 10.1007/978-3-319-13186-3_46 |
收录类别 | EI |
引用统计 | |
文献类型 | 会议论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/11792 |
专题 | 模式识别国家重点实验室_视频内容安全 |
通讯作者 | You Q(游强) |
作者单位 | 中国科学院自动化研究所 |
推荐引用方式 GB/T 7714 | You Q,Hu WM,Wu O,et al. A Unified Fusion Framework for Time-related Rank in Threaded Discussion Communities[C]:Springer International Publishing,2014:pp 513-524. |
条目包含的文件 | 下载所有文件 | |||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
-332129_1_En_46_Chap(437KB) | 会议论文 | 开放获取 | CC BY-NC-SA | 浏览 下载 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论