Uncertain data-oriented probability query quality optimization method

Abstract

The invention discloses an uncertain data-oriented probability query quality optimization method. Finite resources are given; the query quality is measured by utilizing a combination entropy-based quality function; and an uncertain object set needed to be cleaned can be determined, so that the purpose of optimal expected query quality is achieved under specified resource limitations. The method can comprise two major parts including query quality calculation and cleaning object optimization. During the query quality calculation, all possible query result object sets are indexed by adopting ASI, so that the traversal of all possible worlds in the quality calculation process is effectively avoided, the probability of the query result sets can be quickly updated, and the quality calculation efficiency is improved. During the cleaning object selection, two heuristic rules including candidate subsets and quality function monotonicity are utilized, and an accurate cleaning object optimization algorithm and two similar cleaning object optimization algorithms are proposed, so that while the query quality is remarkably improved, the cleaning object optimization time is effectively shortened, and the cleaning cost is ensured to be in a given budget range.

Description

technical field [0001] The invention relates to database query processing technology, in particular to a probability query quality optimization method for uncertain data. Background technique [0002] The generation of uncertain data originates from many real-life applications, such as sensor input noise, wireless transmission errors, data errors and omissions in data integration, etc. Therefore, in the field of databases, the processing of uncertain data queries (such as probabilistic Skyline calculation, probabilistic k-nearest neighbor query, probabilistic Top-k query, etc.) has received extensive attention. [0003] In an uncertain database, the query results returned by general probability queries are objects with non-zero result probabilities. The uncertainty of the data set will be propagated to the query results, so it is difficult for users to get the expected accurate query results. Low-quality query results are also difficult to help users make correct decisions...

AI Technical Summary

Problems solved by technology

[0004] As we all know, data cleaning is an effective way to improve quality, but it is also a time-consuming and costly process, especially in a big data environment, it is impractical to clean all uncertain data

At present, scholars at home and abroad have done some work on the modeling and query processing of uncertain data, but these works still have limitations: (1) the query algorithm does not consider the limited user resources; (2) data cleaning There is no effective algorithm for probabilistic Skyline query and probabilistic k-nearest neighbor query in the algorithm

Images