Support for Document Revision in an Organization Using the Association Rule

摘要

The association rule is applied to revision data attained through the document revision process in an organization, and rules for document revision are extracted. Information characteristic of the organization is intrinsic to this revision data and, therefore, the authors believe that revision support based on organizational information, as well as at the notational and grammatical level, will become possible with the extracted rules. The proposed technique generates revision data from the differences in a document file before and after revision, and then mines revision rules from this data. At this time, an index called adaptive degree is introduced in order that important rules are not overlooked even if the frequency of those revisions is low. Furthermore, using a document which has been completely revised, the adaptive degree is changed according to the domain of the document. The experiment provided the following results relating to the effectiveness of the proposed technique. (1) It was possible to mine rules expressing revisions at the notational, grammatical, and semantic level and revisions based on organizational information. (2) It was presumed from the percentage of terminology included in the rules that the change in the adaptive degree of a rule according to the domain was performed appropriately. (3) Eighty percent of the revisions by a reviser could be made with rules; and the rate of the revising error could be reduced by setting the threshold value of the adaptive degree to an appropriate value.