A New Compression Method with Fast Searching on Large Databases
Jianzhong Li, Doron Rotem, Harry K. T. Wong · eScholarship (California Digital Library) · 1987
In this paper, a new comp~ession method for constant removal from very large scientific and statistical databases is presented.The new method combines the best features from several classical constant removal compression methods.The result, both analytical and experimental, shows that the method is superior to these popular methods in terms of compression effectiveness and efficient searching on the compressed data.In addition to the development, analysis and validation of this new method, this paper.also presents analysis of several traditional constant removal methods for the purpose of analytic comparison.A large collection of experiments have been designed and run to observe and validate the behavior of the compression methods.Another contribution of the paper is that performance characteristics are identified for different compression methods under different data properties assumptions.The result can be used as a basis of selecting compression methods.by matching the properties of the database at hand to the data properties experimented in the paper.