Incomplete relation revision method based on template

Data quality is the core problem in the field of big data application. In practical applications, data are often derived from multi-source databases, which can cause named conflict and similar duplicate record problems. This paper proposed a method to solve the named conflict problem and similar duplicate record based on a given template. In order to find the relationship between the template data and the data to be unified, we segmented the data that to be unified, then built an extensional B tree based on these data. At the same time, we construct a NFA for each template value. By using these NFA, we can get the language pattern for each column in the template relation. Finally, we search each template value in the extensional B tree, if the template value can be found and the corresponding data to be unified can be accepted by the NFA based on the template value, we can use the template value to replace the data that need to be recovered. Then, the data can be consolidated and integrated to ensure the consistency and integrity of the data.