Flooding is the most famous technique for locating contents in unstructured P2P networks. Recently traditional flooding has been replaced by more efficient dynamic query （DQ） and different variants of such algorithm...Flooding is the most famous technique for locating contents in unstructured P2P networks. Recently traditional flooding has been replaced by more efficient dynamic query （DQ） and different variants of such algorithms. Dynamic query is a new flooding technique which could estimate a proper time-to-live （TTL） value for a query flooding by estimating the popularity of the searched files, and retrieve sufficient results under controlled flooding range for reducing network traffic. However, all DQ-like search algorithms are ＂blind＂ so that a large amount of redundant messages are caused. In this paper, we proposed a new search scheme, called Immune Search Scheme （ISS）, to cope with this problem. In ISS, an immune systems inspired concept of similarity-governed clone proliferation and mutation for query message movement is applied. Some assistant strategies, that is, shortcuts creation and peer traveling are incorporated into ISS to develop ＂immune memory＂ for improving search performance, which can make ISS not be blind but heuristic.展开更多
We investigated the application of Causal Bayesian Networks (CBNs) to large data sets in order to predict user intent via internet search prediction. Here, sample data are taken from search engine logs (Excite, Altavi...We investigated the application of Causal Bayesian Networks (CBNs) to large data sets in order to predict user intent via internet search prediction. Here, sample data are taken from search engine logs (Excite, Altavista, and Alltheweb). These logs are parsed and sorted in order to create a data structure that was used to build a CBN. This network is used to predict the next term or terms that the user may be about to search (type). We looked at the application of CBNs, compared with Naive Bays and Bays Net classifiers on very large datasets. To simulate our proposed results, we took a small sample of search data logs to predict intentional query typing. Additionally, problems that arise with the use of such a data structure are addressed individually along with the solutions used and their prediction accuracy and sensitivity.展开更多
There are current, historical and future information about continuously moving spatio temporal objects. And there are correspondingly spatio temporal indexes for current, past and future querying. Among the various ty...There are current, historical and future information about continuously moving spatio temporal objects. And there are correspondingly spatio temporal indexes for current, past and future querying. Among the various types of spatio temporal access methods, no one can support historical and future information querying. The Time Parameterized R tree(TPR tree) employs the idea of parametric bounding rectangles in the R tree. It can effectively support predictive querying to continuously moving objects. Unfortunately, TPR tree can not used to historical querying. This paper presents a partial persistence method in order to extend TPR tree for querying past information of moving objects. In this method, several TPR trees will be created for more effectively predictive querying, because TPR tree has a time horizon limit for predictive querying. Further more, a B tree will be used to index time dimension. Since the partial persistence method brings about huge storage space using, this paper also discusses some methods on how to reduce storage space. Finally, this paper presents an extensive experimental study for the proposed method and gives some interesting directions for future work.展开更多
提出了一种基于概率模型的预测性时空区域查询处理方法.该方法采用Filter-Refinement方式来处理查询.首先,从数据库中选择所有可能满足查询的候选移动对象;然后,根据概率模型中定义的方法来计算候选移动对象满足查询的概率;最后,根据查...提出了一种基于概率模型的预测性时空区域查询处理方法.该方法采用Filter-Refinement方式来处理查询.首先,从数据库中选择所有可能满足查询的候选移动对象;然后,根据概率模型中定义的方法来计算候选移动对象满足查询的概率;最后,根据查询中指定的最小概率阈值过滤候选移动对象并返回查询结果.该概率模型将移动对象未来可能出现的位置定义为一个随机变量,并给出了计算移动对象在两种不同的运动模式下满足查询的概率值的方法.还提出了一种通过对大量历史轨迹抽样来获得概率密度函数(probability density function,简称PDF)的轨迹分析算法,并设计了概率密度函数索引STP-Index(spatio-temporal PDF-index).该索引能够有效地提高轨迹分析算法和概率计算的效率.实验结果表明,该查询处理方法能够有效地支持预测性时空区域查询的处理,提高查询结果的正确性,特别适合于具有较小的空间区域和长时间范围的预测性时空区域查询.展开更多
在目标检索领域,当前主流的解决方案是视觉词典法(Bag of Visual Words,BoVW),然而,传统的BoVW方法具有时间效率低、内存消耗大以及视觉单词同义性和歧义性的问题。针对以上问题,该文提出了一种基于随机化视觉词典组和查询扩展的目标检...在目标检索领域,当前主流的解决方案是视觉词典法(Bag of Visual Words,BoVW),然而,传统的BoVW方法具有时间效率低、内存消耗大以及视觉单词同义性和歧义性的问题。针对以上问题,该文提出了一种基于随机化视觉词典组和查询扩展的目标检索方法。首先,该方法采用精确欧氏位置敏感哈希(Exact Euclidean LocalitySensitive Hashing,E2LSH)对训练图像库的局部特征点进行聚类,生成一组支持动态扩充的随机化视觉词典组;然后,基于这组词典构建视觉词汇分布直方图和索引文件;最后,引入一种查询扩展策略完成目标检索。实验结果表明,与传统方法相比,该文方法有效地增强了目标对象的可区分性,能够较大地提高目标检索精度,同时,对大规模数据库有较好的适用性。展开更多
基金Supported by the National Natural Science Foundation of China (90604012)
文摘Flooding is the most famous technique for locating contents in unstructured P2P networks. Recently traditional flooding has been replaced by more efficient dynamic query （DQ） and different variants of such algorithms. Dynamic query is a new flooding technique which could estimate a proper time-to-live （TTL） value for a query flooding by estimating the popularity of the searched files, and retrieve sufficient results under controlled flooding range for reducing network traffic. However, all DQ-like search algorithms are ＂blind＂ so that a large amount of redundant messages are caused. In this paper, we proposed a new search scheme, called Immune Search Scheme （ISS）, to cope with this problem. In ISS, an immune systems inspired concept of similarity-governed clone proliferation and mutation for query message movement is applied. Some assistant strategies, that is, shortcuts creation and peer traveling are incorporated into ISS to develop ＂immune memory＂ for improving search performance, which can make ISS not be blind but heuristic.
文摘We investigated the application of Causal Bayesian Networks (CBNs) to large data sets in order to predict user intent via internet search prediction. Here, sample data are taken from search engine logs (Excite, Altavista, and Alltheweb). These logs are parsed and sorted in order to create a data structure that was used to build a CBN. This network is used to predict the next term or terms that the user may be about to search (type). We looked at the application of CBNs, compared with Naive Bays and Bays Net classifiers on very large datasets. To simulate our proposed results, we took a small sample of search data logs to predict intentional query typing. Additionally, problems that arise with the use of such a data structure are addressed individually along with the solutions used and their prediction accuracy and sensitivity.
基金This work is supported by the Major Project of National Natural Science Foundation (4 0 2 35 0 5 6 ) andthe Major Project of Natural Science Foundation of Beijing(4 0 110 0 2 )
文摘There are current, historical and future information about continuously moving spatio temporal objects. And there are correspondingly spatio temporal indexes for current, past and future querying. Among the various types of spatio temporal access methods, no one can support historical and future information querying. The Time Parameterized R tree(TPR tree) employs the idea of parametric bounding rectangles in the R tree. It can effectively support predictive querying to continuously moving objects. Unfortunately, TPR tree can not used to historical querying. This paper presents a partial persistence method in order to extend TPR tree for querying past information of moving objects. In this method, several TPR trees will be created for more effectively predictive querying, because TPR tree has a time horizon limit for predictive querying. Further more, a B tree will be used to index time dimension. Since the partial persistence method brings about huge storage space using, this paper also discusses some methods on how to reduce storage space. Finally, this paper presents an extensive experimental study for the proposed method and gives some interesting directions for future work.
文摘提出了一种基于概率模型的预测性时空区域查询处理方法.该方法采用Filter-Refinement方式来处理查询.首先,从数据库中选择所有可能满足查询的候选移动对象;然后,根据概率模型中定义的方法来计算候选移动对象满足查询的概率;最后,根据查询中指定的最小概率阈值过滤候选移动对象并返回查询结果.该概率模型将移动对象未来可能出现的位置定义为一个随机变量,并给出了计算移动对象在两种不同的运动模式下满足查询的概率值的方法.还提出了一种通过对大量历史轨迹抽样来获得概率密度函数(probability density function,简称PDF)的轨迹分析算法,并设计了概率密度函数索引STP-Index(spatio-temporal PDF-index).该索引能够有效地提高轨迹分析算法和概率计算的效率.实验结果表明,该查询处理方法能够有效地支持预测性时空区域查询的处理,提高查询结果的正确性,特别适合于具有较小的空间区域和长时间范围的预测性时空区域查询.
