scispace - formally typeset
Search or ask a question

Showing papers on "Locality-sensitive hashing published in 1978"


Proceedings Article
13 Sep 1978
TL;DR: This work defines virtual hashings which practically independently of the number of such records find in one disk access almost each record of the file, such that several accesses would be needed if the function initially chosen for the file was used.
Abstract: We propose a new type of hashing which we call virtual hashing. In contrast to any known hashing, a virtual hashing may modify its hashing function. Such changes may be performed when collisions arise. A virtual hashing may then find in one disk access, a record such that several accesses would be needed if the function initially chosen for the file was used. We define virtual hashings which practically independently of the number of such records find in one disk access almost each record of the file.

86 citations


Proceedings ArticleDOI
13 Nov 1978
TL;DR: This paper proposes the use of hash coding to guide the global search for nearest neighbor searching and believes that the method is efficient because hashing is easy to implement and fast to execute.
Abstract: All of the existing nearest neighbor searching techniques divide the searching into two parts: A global searching and a local searching. In order to minimize the number of records to be examined, the global searching must be a very sophisticated one such that the initial local searching will be confined to a very small region. This kind of global searching is usually very time-comsuming. In this paper, we propose the use of hash coding to guide the global search. Given a record, we hash it to an address and the initial searching is to be started by searching records stored in that address. We believe that our method is efficient because hashing is easy to implement and fast to execute. In other words, if our method is used, the global searching time will be very short.

8 citations