Wednesday, December 26, 2018

How to clean a memory of translation and not die in the attempt


And we have them in our previous post. To clean the translation memory of one of our largest and oldest customers, we use retrenchment which we import in an Excel worksheet, and chose theologian program to edit the memory. The memory had thousands of segments, and X bench reported us more than sixteen thousand inconsistencies. It was a work of long hours, but we leave airy, we tell them how.
With our return of Excel to view and open in Defoliant memory time to unify the inconsistencies. As the person who made the task quite known customer, in the majority of cases it knew what was best, but always that doubt arose could consult with managers in charge of the project, who, in turn, would consult with the client.

For changes in memory, he had no more than run "Search and replace" (Ctrl + H) in Defoliant, paste the "non-preferred" version and the "favorite" (by calling them somehow, since there was not always an incorrect) in the corresponding boxes, and make the relevant replacements. Although there is the option of replacing everything, in this case it wasn't a way to implement changes that would give much security. Sometimes, the segment that must be modified is composed of two or three words which also appear in the other instance as part of a sentence, and replace everything without analyzing each case can lead to introduce errors.
Like almost all programs, Defoliant also offers options that the capitalization match, make that whole word match and use regular expressions. This last option was us truly useful when it comes to reviewing the figures. As the customer is engaged in the technology industry, many of his writings are littered with figures: dimensions, quantities and capabilities...