Abstract: Abstract: Vulnerability is an important factor to cause information security problem. It is interesting and significant to predict the occurrence of vulnerability. Aiming at the vulnerability data in security vulnerability database, this paper constructs an information security vulnerability ontology based on CWE (Common Weakness Enumeration), which will be taken as vulnerability domain semantics. By applying Apriori algorithm, the association relationship among vulnerabilities in software is analyzed and predicted. In the phase of data preprocessing, by means of the vulnerability semantic knowledge, the support of itemsets is improved and the hidden association rules are found by promoting the concept level of vulnerability data from low to high. In the phase of rule evaluation, by designing rule filters, ADARF and RDARF, according to the user attention, the rules matching the interesting of user are obtained. Experiments on the CNNVD vulnerability database demonstrate the effectiveness of the proposed method.