WorldServer TM update behaviour: is the leverage based on SID or on Entry origin?

In a WorldServer environment where SIDs are enabled, is the TM leveraged based on the Entry Origin or on the SID? For example, if two tasks contain a segment with the same SID, is it possible to have two translations for that same segment?

The answer to the behavior of leveraging is very dependent upon the settings specified in tm.properties. If the tm.properties file contains non-default settings, the results may be different.

With the default settings, the wrong entry may be selected when subsequent leveraging occurs if a segment exists where both the content and the SID match two entries in the TM.
If a SID match is going to be used then the combination of the source content and SID are expected to be unique.
However, if there are two occurrences of the same content requiring translation, it is typical that the context (the text before and after the segment) should be used as a consideration for which TM entry should be selected.

The way to specify this need in the tm.properties file is to have the following lines:

prefer_sid_over_ice_matches = false
require_full_usage_context_for_ice = true

The setting  prefer_sid_over_ice_matches = false  causes the search for a match to be first performed without using the SID. Only if there is no ICE nor 100% match and no SID will a SID match be attempted. For the non-SID matches, a full context is required to have an ICE match.

Please note all property setting changes are global and will impact overall leveraging.



 

This article explains how to configure SIDs in a WorldServer environment:

How to set up a SPICE (SID) Matching configuration in WorldServer