Mode matching method based on implicit classifying information

Abstract

The invention relates to a mode matching method based on implicit classifying information, which belongs to the technical field of databases. The method comprises the following three steps of: finding out the probable classification of source data example; searching classification attribute; and matching relevant corresponding filter condition for the attribute. The invention limits the application range of attribute matching and improves the quality of data exchange; compared with the traditional Context-Based technology, the method searches real classifying information from the data exampleto construct the constraint condition without depending on classification attribute display; and no matter whether classification attribute display exists in the source mode or not, the method in theinvention can extract constraint information from the example data, limit attribute matching and improve the data conversion quality.

Description

technical field [0001] The invention belongs to the field of databases, and in particular relates to mapping and matching technologies between relational patterns in databases. In particular, it relates to a method for discovering implicit classification information to improve the quality of pattern matching. Background technique [0002] As computer technology continues to penetrate into various fields, data has shown characteristics such as distribution, self-control, and heterogeneity. Data integration and data exchange are an effective means to solve the problem of sharing among multiple data sources. The application fields of data integration and exchange are extremely extensive, such as ontology merger, data space, data warehouse, e-commerce, life science, Deep Web and so on. As the preparatory stage of data integration and exchange, the main task of schema matching is to discover the semantic similarity between schema elements. If two schema elements have similar s...

AI Technical Summary

Problems solved by technology

If there are a large number of low-potential attributes, it is likely that the algorithm cannot find display classification attributes

The confusion between low-potential attributes and display classification attributes makes the algorithm unable to distinguish them, resulting in lower search accuracy

[0006] (2) Second, the return result of the first step of the technology is often some candidate display classification attributes. Candidate constraints are constructed through these attributes to further form candidate attribute matches, and then the attribute match with the highest confidence is selected, which makes the target Schemas have attributes that match from multiple source schemas, which is often not true

At the same time the candidate space is often large, resulting in a very time- and resource-intensive search for a match with the highest confidence

At the same time, this method is very dependent on the display classification attribute. If the source schema does not have the display classification attribute, the Context-Based technology will fail

Images