With each segment's translation Beebox stores a wide variety of attributes. One of these is the "Origin of translation".
It stores who is at the origin of the current translation. The origin has a title used in the UI and a code which is used in XLIFF and JSON exports. In rare cases we also use a numeric value (in some API calls).
The possible values are listed below:
The translation was last edited by a human.
The translation comes from an MT system and was not post-edited by a human. Also look at the status, it may be that a human set it to “green” to approve the MT result as perfect.
AQA algorithm in the translation management system modified the translation. For example, mark up might have been found badly placed and was fixed by the algorithm. Note that such automatic fixes may not be perfect. Also look at the status of the translation. Was it set to “green” color by a human?
The identical source text was found in a translation management system or a Beebox translation memory and the translation of that hit was used. This means that the translation likely is ok.
The identical source text was found in a TMS or a Beebox translation memory and the translation of that hit was used. In addition the text before and after the source text perfectly matches the text before and after the hit. This means that the translation is “in context” and can typically be highly trusted.
A similar source text was found in a TMS translation memory and the translation of that hit was used. You should look at the status of the translation and see whether a human set it to “green” to approve this leveraging. Note that leveraging inside the Beebox looks for identical texts only.
The translation was leveraged from the previous file version. The context, i.e. the preceeding and trailing source texts, changed. You will likely see this value when a new version of a file was added to the project folder with new pieces of text added.
The translation was leveraged from the previous file version and the context did not change. These translations can be highly trusted.