Translation Memory Search in SDL Managed Translation

Dear Community,

I'm Remy Jacobi and I'm part of the user experience design team, responsible for the user experience around everything translation management. In this function, I regularly engage with our customers and users to better understand their day to day tasks and problems. This ongoing research allows me to design relevant and delightful user interfaces and experiences.

We’re constantly looking for ways to improve the user’s experience in Managed Translation and deliver new features that are being requested by our customers.

One feature we are regularly asked about is the ability to perform a search against a Translation Memory in Managed Translation. Translation Memories help promote quality and consistency, whilst reducing costs and turnaround times for translation projects.

The Translation Memory Search feature will allow users to browse the contents of their translation memory(s) in order to:

  • Look for segments that are already present in a Translation Memory
  • Adapt the source content according to segments already present in a Translation Memory
  • Identify budgetary project costs
  • Determine the language(s) in which segments have been previously localised 

Provided below is a mock-up to help visualise how this feature will be presented in Managed Translation. In the left side of the mock-up, the user is checking to see if the segment: “Insert the cartridge in the left socket of the printer’s box” is already present in one of their Translation Memories. On the right side, a list of results is populated from highest match to lowest.

Exact match implies the segment was found in the translation memory without any penalties having been applied. Matches can be penalised based on a number of factors (text changes, order of words, punctuation, formatting, etc.) Penalties can also be applied at the Translation Memory level, where it contains legacy content for example. Both types of penalties are taken into account in the search results returned by the Translation Memory Search feature.

Would the Translation Memory search be a useful feature for you?

Do you find it easy to understand/read the below mock-up?

We’d appreciate any feedback or comments as to how the Translation Memory Search feature would benefit your organisation along with any suggestions on the proposed mock-ups included in this Community post. :)

Parents
  • Hi Remy, a couple of things: will there be an option to edit segments directly in the TM? For me it's important for several reasons, but mostly to help SDL to be able to update TMs easily in case global changes are needed and also to give the chance to some users on the client side to make changes as needed. Obviously this should be a select group of users, but I see a lot of value in it. Also, will there be a column that specifies the user that last changed or touched the string? One more, can you search throughout the whole TM group including all TMs? It'd be important to be able to do so and then user can sort results based on where the string comes from (sort by TM name). Finally, this is a tool that is much needed and I hope it gets implemented soon, I'm excited about it and can't wait to start using it!
  • Hi Martiño,

    Thanks for taking the time to look at the mock-up and for posting your comments! I'm happy you found the TM search useful for you.
    To answer your questions:

    We won't be supporting segments edits in the initial release of this feature but it's something we will consider as a future enhancement. The editing capability would be controlled through additional permissions.
    Created/Modified By fields are not part of the current design, primarily down to the way that information is captured in the TM format we are working with.
    When performing a search, we look through all TMs present in the selected Project Option, applying a ‘default’ TM penalty where appropriate. If there is a requirement to search through all TMs, a project option could be created/adapted to facilitate this.

    I hope this was helpful.
    Thanks again for your feedback! We will keep you updated about any changes.
  • Hi Rémy and , two more pieces of feedback about the tool which is really making my life better by the way! It would be really useful to have more metadata about each of the strings, for example what project is it coming from, file name, and as of now it looks like the information about Created By and Last Updated By is always “serviceorion” instead of the actual user. Also, something that didn't think about before but I see as an important piece is to be able to look by target strings and not just by source. Let me know if this is the right place for this feedback or want me to create a new post!
  • Hi Martiño, I'm happy to hear that you're finding good use of the TM search.

    Having more metadata would help you understand the source of the TM. How about the current metadata? Which of the values are you constantly looking at when searching for TM?
    I've already asked about the name "serviceorion" that's being populated in those two labels you mentioned, I'll get back to you when I have any update.
    It makes complete sense to search from both, source and target strings. I'm interested in knowing how you would use it. Would you in particular, search in the target string for a specific language that you understand too?

    Thanks for your feedback, and whenever you have other input, please don't hesitate to share it :)
  • Hi Martiño,

    Thanks for sharing your feedback on the TM Search feature - it's great to hear this is making life easier for you!

    When a TM segment is updated 'systematically' as part of a workflow step ("Update TM"), we set the user details (Created By, Last Updated By) to the system name - "serviceorion" in your case. For segments that are created or modified outside of this workflow, via a standalone maintenance tool, we store the details of the individual user. I will check what it would take to record details of the individual user for the automated flow - this won't be trivial, I don't suspect, but I'll at least make sure we consider this as a future enhancement.

    The other metadata you refer to (Filename, Project ID/Name) should be a little more straightforward to implement. I would appreciate it if you could submit this as a new enhancement request via the Ideas section on the Community so we can plan to implement this in a future product update.

    Per the earlier question from Remy, can you please describe how a search-for-target-string feature would be used, and by whom? Is that a different use case to how the TM Search feature is being used for source strings?

    Cheers,
    Ian
  • Hi Ian, it would be really useful to know the last actual person/user that modified the segment more than just the generic workflow name, for multiple reasons. I´ll submit the metadata request via Ideas. Regarding the search-for-target, it is basically the exact same logic as searching by source, sometimes we have discussions with regional offices about content in their own language and we just want to know what the source was or if it's even a translation and it's in our TM vs content created locally, etc. It's helpful in a few scenarios, specially when dealing with regional stakeholders that will talk about content in their language vs thinking about the source.
Reply
  • Hi Ian, it would be really useful to know the last actual person/user that modified the segment more than just the generic workflow name, for multiple reasons. I´ll submit the metadata request via Ideas. Regarding the search-for-target, it is basically the exact same logic as searching by source, sometimes we have discussions with regional offices about content in their own language and we just want to know what the source was or if it's even a translation and it's in our TM vs content created locally, etc. It's helpful in a few scenarios, specially when dealing with regional stakeholders that will talk about content in their language vs thinking about the source.
Children
No Data