Extracting relations among security patterns

The activity of the secure system development can be supported by reusing extensive knowledge accumulated about security in the form of security patterns. There are a number of catalogs of security patterns available on WWW and literatures; however all of relations among security pat-terns are closed in each pattern catalog. Moreover even in each catalog, the author of the catalog might have over-looked useful relations among patterns belonging to the same catalog. This situation makes the selection and appli-cation of the right pattern for each security development ac-tivity a daunting task. To acquire such useful but overlooked relations in each catalog and cross-cutting relations over different catalogs, we have applied our technique for the automatic pattern relation analysis to a set of security pat-terns. Our technique utilizes existing text processing tech-niques to extract patterns from documents and to calculate the strength of pattern relations. As a result of experimen-tal evaluations, it is found that our technique can extract appropriate relations in each security pattern catalog and over different catalogs, without information on relations de-scribed in original pattern documents. These newly found relations will be useful for retrieving, selecting, and com-bining security patterns.