scholarly journals Fast and Exact Nearest Neighbor Search in Hamming Space on Full-Text Search Engines

Author(s):  
Cun (Matthew) Mu ◽  
Jun (Raymond) Zhao ◽  
Guang Yang ◽  
Binwei Yang ◽  
Zheng (John) Yan
Author(s):  
Zhansheng Jiang ◽  
Lingxi Xie ◽  
Xiaotie Deng ◽  
Weiwei Xu ◽  
Jingdong Wang

2014 ◽  
Vol 651-653 ◽  
pp. 2168-2171
Author(s):  
Qin Zhen Guo ◽  
Zhi Zeng ◽  
Shu Wu Zhang ◽  
Yuan Zhang ◽  
Gui Xuan Zhang

Hashing which maps data into binary codes in Hamming space has attracted more and more attentions for approximate nearest neighbor search due to its high efficiency and reduced storage cost. K-means hashing (KH) is a novel hashing method which firstly quantizes the data by codewords and then uses the indices of codewords to encode the data. However, in KH, only the codewords are updated to minimize the quantization error and affinity error while the indices of codewords remain the same after they are initialized. In this paper, we propose an optimized k-means hashing (OKH) method to encode data by binary codes. In our method, we simultaneously optimize the codewords and the indices of them to minimize the quantization error and the affinity error. Our OKH method can find both the optimal codewords and the optiaml indices, and the resulting binary codes in Hamming space can better preserve the original neighborhood structure of the data. Besides, OKH can further be generalized to a product space. Extensive experiments have verified the superiority of OKH over KH and other state-of-the-art hashing methods.


Author(s):  
Pavel Šimek ◽  
Jiří Vaněk ◽  
Jan Jarolímek

The majority of Internet users use the global network to search for different information using fulltext search engines such as Google, Yahoo!, or Seznam. The web presentation operators are trying, with the help of different optimization techniques, to get to the top places in the results of fulltext search engines. Right there is a great importance of Search Engine Optimization and Search Engine Marketing, because normal users usually try links only on the first few pages of the fulltext search engines results on certain keywords and in catalogs they use primarily hierarchically higher placed links in each category. Key to success is the application of optimization methods which deal with the issue of keywords, structure and quality of content, domain names, individual sites and quantity and reliability of backward links. The process is demanding, long-lasting and without a guaranteed outcome. A website operator without advanced analytical tools do not identify the contribution of individual documents from which the entire web site consists. If the web presentation operators want to have an overview of their documents and web site in global, it is appropriate to quantify these positions in a specific way, depending on specific key words. For this purpose serves the quantification of competitive value of documents, which consequently sets global competitive value of a web site. Quantification of competitive values is performed on a specific full-text search engine. For each full-text search engine can be and often are, different results. According to published reports of ClickZ agency or Market Share is according to the number of searches by English-speaking users most widely used Google search engine, which has a market share of more than 80%. The whole procedure of quantification of competitive values is common, however, the initial step which is the analysis of keywords depends on a choice of the fulltext search engine.


Author(s):  
Bilegsaikhan Naidan ◽  
Magnus Lie Hetland

This article presents a new approximate index structure, the Bregman hyperplane tree, for indexing the Bregman divergence, aiming to decrease the number of distance computations required at query processing time, by sacrificing some accuracy in the result. The experimental results on various high-dimensional data sets demonstrate that the proposed index structure performs comparably to the state-of-the-art Bregman ball tree in terms of search performance and result quality. Moreover, this method results in a speedup of well over an order of magnitude for index construction. The authors also apply their space partitioning principle to the Bregman ball tree and obtain a new index structure for exact nearest neighbor search that is faster to build and a slightly slower at query processing than the original.


2020 ◽  
Vol 99 ◽  
pp. 107082 ◽  
Author(s):  
Bin Fan ◽  
Qingqun Kong ◽  
Baoqian Zhang ◽  
Hongmin Liu ◽  
Chunhong Pan ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document