The daunting task of cleaning a translation memory


A few months ago we decided to clean the translation memory of one of our clients with more workload and more long-standing.Over many years of work, the memory was loading inconsistencies and mistakes that confuse the translator when he does his work.
The fact that no errors and inconsistencies is not that translators are little professional, but characteristics of the text and changes to the preferences of the customer over time. Times change, changing contexts and changes of course, also the most convenient way to translate a specific text. For this reason, once in a while, should clean the memory to not continue to drag outdated segments that no longer serve us.
The first step, knowing what had to be corrected, it was simple. We used Bench, a tool with which we always work to make our quality controls. We carry memory in X bench and extrude a report of inconsistencies in the text source, inconsistencies in the text meta and spelling errors. The latter were very few. the majority were false positives. But more than sixteen thousand inconsistencies made us think that we were facing a never ending task. To make it easier to move between errors and leave comments about any doubts or queries, we import the Benchley report in an Excel file.
However, what seemed more complicated was managing thousands of entries that were part of our memory and modify them without generating even more errors. The possibilities offered us the language translation services  tools we use do not allow us to manage memory as we needed, we decided to use Defoliant, a program that allows open memories format .tux or .text and edit them. Something to note is that by default, Defoliant looks for in the source text, but we can choose to look only in the text source, the target text or several other options. Just check what you want to do.
He finally began the daunting task of delete, change, correct... The instructions were pretty simple and, if you will, obvious: unify the inconsistencies and correct any spelling. But there was to eliminate any segment as each one of the segments that make up the memory is related to another that follows it or above it, and delete one alters that relationship and, therefore, the memory.

Comments

  1. Buy Yahoo phone verified accounts
    https://instantboostup.com/pva-accounts/buy-yahoo-phone-verified-accounts/

    ReplyDelete

Post a Comment