Customers often ask how translation memory is organized. Is it stored in one master database or in project databases? The first step is to define the different ways to organize translation memory.
Master translation memory – this is where you place all translation memory for a specific client into a master database. This would mean projects from all divisions or departments and multiple types of document formats (Word, Flare, FrameMaker, etc.). The content in the database carries meta tags for project name, product name, domain, date created, etc. The meta information can be useful if you want to create a subset of the database.
Project translation memory – this is where you limit the translation memory to a given project. This can be useful if you are trying to isolate translation memory for a specific project. If the meta information is handled properly in the master database you can always export a subset from the master memory.
So which one is better? Generally speaking, properly maintaining a master translation memory is the best practice to follow. The key is properly organizing the memory with meta information that would allow you to split the data into meaningful pieces if required.
Keeping project memories works well for the first few projects but what happens when you complete the 100th assignment? Managing that many project memories for one client is very difficult. You will certainly forget to utilize one or ten (!) relevant translation memories for a subsequent project. The goal of using translation memory is to maximize the amount of content you can reuse from project-to-project which will in turn lower your cost, decrease the time required and increase consistency.
It is best to keep your approach simple!