Top-K representative documents query over geo-textual data stream
Top-K representative documents query over geo-textual data stream
复制标题
地理文本数据流的Top-K代表性文档查询
DOI：
10.1007/s11280-017-0470-0
复制
发表时间：
2018
影响因子：
3.7
通讯作者：
Wang Guoren
中科院分区：
文献类型：
--
作者：
Wang Bin;Zhu Rui;Yang Xiaochun;Wang Guoren
基于位置的社交网络的日益普及鼓励越来越多的用户分享他们的经验。本文研究了地理文本数据流上的高价值文档查询问题。许多研究人员已经研究了这个问题。但是，他们没有考虑文件的可靠性，一些不可靠的文件可能会误导客户做出不正确的决定。此外，它们缺乏修剪具有低代表性的文档的能力。为了提高推荐系统的用户满意度，我们提出了一个新的框架PDS。它首先采用一种有效的机器学习技术ELM来修剪不可靠的文档，然后使用一种新的索引来维护文档。首先，这个索引维护了一组剪枝值来过滤低质量的文档。另一方面，它利用滑动窗口的独特性质，进一步提高了PDS的性能。理论分析和大量的实验结果证明了所提出的算法的有效性。
The increasing popularity of location-based social networks encourages more and more users to share their experiences. It deeply impacts the decision of customers when shopping, traveling, and so on. This paper studies the problem of top-Kvaluable documents query over geo-textual data stream. Many researchers have studied this problem. However, they do not consider the reliability of documents, where some unreliable documents may mislead customers to make improper decisions. In addition, they lack the ability to prune documents with low representativeness. In order to increase user satisfaction in recommendation systems, we propose a novel framework namedPDS. It first employs an efficiently machine learning technique namedELMto prune unreliable documents, and then uses a novel index namedto maintain documents. For one thing, this index maintains a group ofpruning valuesto filter low quality documents. For another, it utilizes the unique property of sliding window to further enhance thePDSperformance. Theoretical analysis and extensive experimental results demonstrate the effectiveness of the proposed algorithms.