Documentation Center

Examples of Using Quality Models

The most common use case for Quality Models is one where an LSP or company receives a “finished” translation and then decides to perform a quality audit on this work. This could involve taking a random sampling of the assets returned from the translator and sending them out to a separate reviewer to perform a quality audit. This auditing reviewer could be a subject matter expert. The reviewer would go through and audit the translation by flagging each segment with errors. The reviewer would return the audited asset to the company or LSP. The asset could then be “scored” based on the types of errors in the document. Based on this score, the company could decide whether to keep the translation or send it back to the translator. In addition, this quality information would be preserved for future reporting purposes.

A second use case is one in which the translators and reviewers use this error information during their regular review process to provide more specific information about why they rejected a segment. This use case can really be thought of as an extension of the translation review cycle described in Segment Status in a Translation and Review Workflow; the reviewer simply has some additional information they can use when rejecting a segment.

The primary difference between this use case and the previous one is the end result of the segments. In this second use case, the most likely end state of the asset is that all of the segments will be cleared of their errors. Specifically, by the time the asset gets through the review process, all of the errors should have been fixed. Because of this, this use case is not really conducive to historical reporting—the error information is transient and goes away by the end of the project.