Overview of TM AIS path normalization
During the storage of translation pairs, the TM entries are associated with the TM AIS context of the asset from which they were created. Each asset effectively has its own set of translations within the TM. If the asset is updated and retranslated, the new translations will replace the previously stored translations that are associated with the asset. This allows WorldServer to maintain separate translation sets for similar or identical content shared across various assets. For example, if two assets with identical content are translated differently, the TM will contain a separate set of translations for each asset. When re-leveraging an asset against a TM, WorldServer will prefer TM entries that were created from an earlier version of that asset, even if there are newer TM entries from another asset for the same content. (See the section on WorldServer Rules for Ranking Matches.) This is important in that it allows WorldServer to regenerate target files long after they were originally created.
There have been occasions where customers have desired that the TM AIS context be shared between multiple assets. For example, a customer may be working on different versions of the same file, and the different versions may be stored at different locations within the AIS mounts. When translated, each of these versions will have their own set of translations in that they are treated as separate assets.
Path normalization is the process through which the TM AIS context can be modified in order to support the sharing of the TM AIS context between multiple assets. TM AIS context, similar to the AIS context, is based on the AIS path of the asset being translated. The path normalization facilities allow you to augment the TM’s definition of AIS path. For example, AIS path differences related to version differences can be normalized out of the AIS paths.
/AIS/VERSION1/EN/PRODUCT1/introduction.doc
/AIS/VERSION2/EN/PRODUCT1/introduction.docThe above referenced assets may contain different versions. WorldServer will treat these as two separate assets because the AIS paths are different and because WorldServer does not require asset names to be unique. If the customer wanted these two assets to share the same TM translations, then the path normalization process would need to be overridden to negate the differences between the two paths. One normalization strategy might be to simply remove the version element from the two paths:
/AIS/EN/PRODUCT1/introduction.doc
/AIS/EN/PRODUCT1/introduction.doc
Now, the two paths will look identical when presented to the TM storage level, and the TM will see both assets as being the same asset.
- They share a single set of TM translations.
- The translations in the TM will match those of the most recently translated version.
- Potential loss of translation regeneration capability of all assets. If particular versions of the assets sharing the TM AIS context are not identical in content, the TM entries will only completely match the segments of the last translated version. This means that it is possible to lose some of the associated translations or that some ICE matches may become merely exact matches due to potential content and segmentation differences between the different versions. Re-leveraging a previously translated version may no longer be completely ICE matched or translatable against the TM.
Updates to one asset will affect the translations for all other assets sharing the TM AIS context when they are re-leveraged.