Abstract: To effectively clean the dirty data among the database,a variety of integrity constraints have been proposed,such as Conditional Functional Dcpcndcncics(CFD) , Conditional Inclusion Dcpcndcncics(CIDS). Even though these constrains are compentent to detect the existence of mistakes,they couldn't effectively guide us to corretc these mistakes,as a mater of fact, data repairing based on these constraints maybe not able to find certain fixs that arc absolutely right,what's more, thay may introduce new mistakes, so it reduced the effenciency of data repairing. Focusing on the above-mentioned demerits, this paper proposed a better data repairing framework; firstly, those fixes which arc based on Editing Rules and Master Data arc bound to be certain, we also provide an an algorithm to automatcly repairing the dirty data; seconldly, the prior step may not repair the whole attributes of the relation, so we employ the CFDto correct the reamining dirty data, unfortunatly, these fixes arc possible fixes which maybe not totally right. Even so, compared with others, the framework show great superiority, not only enhance the efficiency and uniqueness, but slao make sure the percision of data repairing.

Key words: Conditional functional dependency, Cleaning rules, Data cleaning, Data quality