Although publishing views supply convenience to data exchange, it will potentially threat data owner because some information may be disclosed in the publishing process, so guaranteeing the security of the publishing views becomes a new issue of database security. Theoretically, the methods preventing the information from disclosing in the publishing process can be divided into two kinds, one is for the people who accept views, and the other is for the people who publish views. In practical applications, it is very difficult to realize the first method, so, people focus study on the second method. So far, relevant evaluating algorithms and protecting models have been proposed, for example, query answering, K-anonymity, probabilistic independence event model and so on, but all these methods have the same weakness, namely, limitation and can't eliminate information disclosure effectively, so they cannot resolve the problem completely. In order to measure the leakage of the publishing views and eliminate disclosure, an information disclosure measuring method of relative deviation is proposed, which can facilitate the information disclosure calculation, and a disclosure elimination method based on critical tuples is presented, which can eliminate the information disclosure of the publishing views completely. The experiment results show that the information disclosure measuring method of relative deviation can measure information disclosure effectively, and the information disclosure method based on critical tuples can eliminate information disclosure effectively. They can resolve the problem of information disclosure in the publishing process completely, and can guarantee the security of views in the logic level.