Embodiments of the invention disclose multi-source heterogeneous vulnerability information duplicate removal and grading methods and apparatuses. In the vulnerability information duplicate removal process, firstly weight values of extracted feature words in whole vulnerability information contents are calculated; secondly according to calculated weight value results, the feature words with the weight values sorted in front are selected as effective feature words of vulnerability information; and finally the vulnerability information which the effective feature words can be matched with is classified and grouped to realize vulnerability information duplicate removal. According to the duplicate removal method provided by the invention, vector space dimension calculation is not required, so that the calculation complexity can be lowered; and moreover, the feature words with the weight values sorted in front are screened out to serve as the effective feature words, and a few interference feature words are removed, so that the accuracy of a correlation judgment result can be improved. Likewise, in the information grading process, firstly the feature words of the vulnerability information are extracted; secondly the extracted feature words are subjected to expression analysis of emotional semantics; and through the analysis, the risk levels of the vulnerability information can be obtained.