Reverse leveraging penalty in 11.1.0.3909

Hello!

 

We are trying to get the reverse leveraging penalty enabled according to the information in the Admin Guide V11.1.0. Here is the relevant passage:

 

A configurable penalty is assessed to control auto-leverage.
The tm_score_reverse_leverage_penalty TM property in
tm.properties can be configured to penalize reverse leveraged matches. The
default value is 0. The allowed values must be in the range of 0 to 1. This penalty
is a leverage level penalty, and as such, it is applied to the entire match. (Many
penalties are applied to the segment element—such as a word or number in the
segment—that causes the penalty.) As a result, the impact of the penalty is not
affected by the length of the segment. For example, if the penalty is set to 0.02,
then the final score will be reduced by 2 percentage points. Based on this value, the
maximum score that a reverse leverage match could generate is a 98% score.
The penalty would be applied to all matches resulting from the TM reverse
leverage process. However, it will initially only apply to potential 100% and
fuzzy matches since the TM reverse leverage process will not initially generate
SPICE or ICE matches.

 

I have set the above parameter to 0.02 in all of the tm.properties I can find (webapps\ws, \ws-api, \ws-legacy), restarted the services for WorldServer and the FTS but it's just not working. I've started with a completely blank memory (reverse leveraging naturally enabled), added some TUs in the direction German->English, and then used these English translations to test reverse leverage in another project. Reverse leveraging is working fine, but it is delivering 100% hits instead of 98%. Checking the units from BW also shows scores of 100%.

Can anyone figure out what I'm missing? Thanks for any help!