Documentation Center

Hyper-Merge

Hyper-merging evaluates merge candidates based on language content alone. Placeholders are ignored initially, but are still used to evaluate final scoring of merged segments. Depending on the assigned placeholder penalties, the final segment may result in additional 100% matches, but most probably will lead to higher fuzzy matches only.

For example, consider the earlier scenario where the first segment has been modified, as shown in the following figure.

Next, seed the TM with the following TM entry (which is the same entry as we used earlier):

Even though the translation has not been populated in the target segment, the source segments have still been merged. Clicking the TM entries button shows the following match.

In this example, the final score is a 100% match due to the safe placeholder repair. However, often the score for the hyper merged match is less than 100%. This is still beneficial in that instead of two segments with low fuzzy matches, there is a single segment that yields a high fuzzy match. This result has significant implications to final translation cost that are geared toward varying fuzzy ranges.

The above example shows two segments being merged together. The hyper-merge process can merge an indefinite number of segments to match up to a super TM entry, provided that no complete matches are found along the way.

1 100% matches generally occur only when the placeholder penalties have been set to zero.