Inter-language Vector Space (ILVS) is an NLP AI framework we have developed which improves both translation efficiency and quality. It works out semantic relationships between words across 250 language pairs, and helps localization managers and linguists by automating repetitive, time-consuming tasks.

Erroneous MT translations can be frustrating. This usually happens when several translations are possible or with homonymous terminology. Artificial intelligence can solve this issue by enabling MT results to take the existing TM entries into account. This ensures that the right words are used in the right context by filling in the missing words in a segment from a previous TM entry, converting a low fuzzy match to a high fuzzy or even a 100% match. Thanks to this AI tool, we estimate that the decrease in human input required to amend MT translations can go down on average to approximately 30%. In other words, about 60 hours per month.

Having to place inline tags manually is one of the most time-consuming tasks a linguist has to do. XTM Workbench users can rely on XTM AI to automatically transfer them to their exact location in the target segment. This enhances the linguists’ productivity thanks to only having to focus on the task at hand. With a 98% success rate, XTM’s auto-inline AI tool can save your team up to 200 hours per month. 

Building a TM from scratch can take months of work. To manually build a standard-size TM with 50,000 segments would take approximately 50 hours – XTM’s artificial intelligence is able to do that in minutes. XTM AI matches your legacy source documents with target texts to build a rich translation memory. This not only accelerates your translation times, but it also reduces costs by at least 50%.

XTM Cloud supports leading machine translation (MT) engines which include NMT and customized engines. Users can send their content to an MT provider from our localization hub, enabling streamlined machine translations that save time. Therefore, linguists can focus on tasks that create value, thanks to the NMT taking care of the bulk of the work.

Efficient localization requires being able to retrieve TM matches in the most productive way. With higher leverage comes faster translations and less costs. At XTM, we have developed our own proprietary algorithm for TM leveraging: Weighted Token Levenshtein (WTL). WTL is able to spot matches which normally wouldn’t be recognized despite being almost identical. It can transform 50% matches into 80% ones so that they can be leveraged by the TMS, retrieving more matches than any other TMS. This means that linguists don’t have to translate segments from scratch, which can save them on average 250 hours per month.