Hercules Dalianis & Eduard Hovy
Abstract
This paper describes a procedure to merge two STEP/AP EXPRESS schemata, using automated methods to assist the merging process. The desire to integrate or merge two schemata or ontologies is very common today (for example, it may be necessary to integrate two models describing car manufacturing from two different car manufacturing companies who just merged and need a common car manufacturing model). This task is usually carried out manually and can be very time-consuming. By (semi-)automating the process, the speed and accuracy of the integration process can be increased. We have developed a method that partially automates the process of schema integration, by using various matching algorithms. Each algorithm proposes a set of cross-schema links, sorted by expected reliability, which can then be validated by a human. The algorithms employ concept names, super- and subclass taxonomizations, attributes, and natural language text definitions. The methods are implemented as a set of Perl Programs.