Our E-Journals are cataloged in Roger, with complete bibliographic information about the title, URL, coverage, subject, etc. Instead of duplicating data entry by rekeying them into our Sage database, we automatically import the records into Sage from Roger. Each E-Journal record is retrieved from Roger in XML format. A Java application creates a record for each URL in the Roger record (Roger records can have multiple URLs, but Sage records cannot). The record is then added to the Sage database, matching Roger subjects and headings with Sage subjects and types and setting default values for some Sage fields.

Since there are over 7,000 E-Journal records in Sage, and, like the rest of the web, the E-Journal sites change frequently, we update the E-Journal records on an ongoing basis. Each week, a list of the E-Journal records that have changed in Roger is generated, and these records are reimported into Sage. Because each Roger record can correspond to multiple Sage records, and the fields that differentiate the multiple records - URL and coverage - are the most likely to change, the importing program uses a series of heuristics to match the records to the Sage records. This process successfully matches more than 99% of records, leaving a handful of records to be manually correlated.

Once an E-Journal record has been imported into Sage, our authors can then add Sage-specific information to the resource (such as a description, additional subject and types, or keywords). The basic information about the E-Journal (title, URL, list of subject headings which are placed in a text field for the spider to see) are overwritten when it is updated. But most fields, including subjects, types, and description are left to our authors to modify as they see fit.