Reducing the retrieval time of hashing method by using predictors

Many methods for resolving collisions in hashing techniques have been proposed. They are classified into two main categories: open addressing and chaining. In this paper, other methods are presented that are intermediate between the two categories. The basic idea of our methods is the use of one or more predictors reserved per cell instead of a link field as in the chaining method. The predictors are used to maintain loose synonym chains. After describing the methods, the efficiencies are estimated theoretically and verified experimentally. In comparison with the chaining method, we prove that our methods significantly reduce the average number of probes necessary to retrieve a key without expending extra space.

View this article in the ACM Digital Library.