Under Community Review

MultTerm Extract improvements

It is not possible to perform an extraction “out of the box.” For example, If wish to  perform a bilingual extraction from a TMX file, I cannot do so without having an existing MultiTerm in to which I can import the results. To make matters worse, if I wish to avail myself of the user fields provided in MultiTerm Extract such as Definition, Note Synonyms, and Context Sentences, I will not be able to export these fields values, unless my existing termbase conforms exactly to this structure. If I have an existing termbase with only one user-defined field, all the field values I provided in MultiTerm Extract will be lost.

I believe a basic requirement for MultiTerm Extract would be the ability to export the extraction results to a generic XML file (and Excel spreadsheet), and to provide an .xdt file containing all the fields offered in the program.

After all the years that have passed since MultiTerm Extract was launched, it would also be nice to offer support for SDLXLIFF files and finally fix the issues concerning the learning termbase.

Thank you.

  • IMHO, we need far more than that. We need a modern-era terminology extraction, which could be based on neural networks instead of just plain statistical approach. Also, the SDL Language Cloud terminology more robust and future-proof than MultiTerm, so any improvements in the area of terminology extraction should rather be based on this solution.