We need to be able to assign different QA models within a configuration based on Word Count thresholds.
However only one QA model can be listed per configuration. Incredibly, I was told PS could not even route these to an assigned QA model based on word count at preprocessing.
It was suggested we try a percentage model, however 1) this is extremely complex to set up based on word count thresholds and 2) certain errors should always fail a job, i.e. critical mistranslations, where others may fail job in different numbers, e.g. x number typos fail small job while y number typos fail a larger job so percentage doesn't work well for that.
We should be able to have separate QA models and route according to the WC thresholds or other parameters we may require. Currently the limitation forces us to manage scoring of other thresholds manually offline.