• State Verified Answer
  • Replies 3 replies
  • Subscribers 389 subscribers
  • Views 1633 views
  • Users 0 members are here
Options
Related

RecomputeFuzzyIndexStatistics method for file-based TMs?

Daniel García Magariños

Dear all,

I am a bit confused with the use of the RecomputeFuzzyIndexStatistics method in the FileBasedTranslationMemory class.

According to the SDK, this method "Synchronously recomputes the fuzzy index statistics of the TM." There is a sample for file based TMs in which the following description is included:

"Probably the most common tuning operation is the recomputation of the fuzzy index. As the TM grows more and more units get added to the database. To keep the fuzzy index lean and efficient, you should recompute the index at regular intervals, depending on how fast your TM grows. This ensures speed and accuracy of the TM. Through the API you can determine when the fuzzy index was last recomputed and whether recomputing is required / recommended. Based on this information you could then trigger an action to recompute the fuzzy index programmatically."

Does this description above applies also to file-based TMs?

I am asking because, according to the SDL Studio 2014 SP2 online help, when discussing the "Recompute Fuzzy index statistics" button, it is said: "This option does not apply to file-based translation memories. Fuzzy index statistics are not kept for file-based translation memories."

I was thinking of making a small script that goes through all our file based translation memories (and we quite a few) and then use the  "RecomputeFuzzyIndexStatistics" method on then, hoping to improve the performance.

Will I achieve anything with it with file based TMs? Does it actually make any change on a file based translation memory or will the TM be the same as before?

Thanks!

Daniel
Rate translation
×
Suggest better translation
×
Parents Reply
  • 0 in reply to Romulus Crisan (former member)

    Thanks a lot, Romulus, for your feedback. It is good to know that recomputing the fuzzy index works with file based TMs as well (ours are bigger than 50,000 TUs).

    If the Recomputing the fuzzy index does actually have an effect on file based translation memories, then the user documentation should probably be updated. As it is now, it seems as if Recomputing the fuzzy index did in file based TMs.

    Thanks also for the pointer to Reindex TM. I had seen it and making the source code availble is tremendously useful. I did not think it would be useful in my case because it seems to be do a "Reindex" and not a Re-compute of the fuzzy index statistics. I wil have a look at the code, though.

    As a matter of fact, in our TMs we do not normally change the Recognise settings, which means that the Reindex would have no effect. Or am I missing something?

    Daniel
Children