Segments

In the context of ModelStudio, Segments are source and target lines. Each line is a segment. A segment can contain several sentences, but should not be too long (as too long sentences will be filtered out during the training). Depending on the language pair, this threshold varies, but we recommend avoiding sentences longer than 100 words.

Please refer to the following sections for more information on how to:

• Browse through segments

• Edit segments

• Delete segments

• Add segments

Browse segments

To view segments in a corpus, click on the corpus name. By default, 25 segment pairs are displayed. You can increase this value by clicking on the box in the bottom left corner and choosing another value from the dropwdown list.

You can search through the segments by entering words in the Source and/or Target search fields.

All segments matching your research will be displayed. The matching process is not case-sensitive. For instance, in the example above, sentences containing the words “Card”, “Cards”, “card” are matched by the search term “card”.

Edit segments

All the segments can be modified. Click on a segment to edit it.

Clicking on the gray circled cross on the right end of the edition field will delete the whole text.

Once you’re done editing the segment, click on the check mark button to save your modifications. Click on the cross button to cancel your modifications.

Delete segments

To remove segments from a corpus, tick the boxes left to the segment pairs you want to delete, then click on the Delete button.

You will be asked to confirm your decision to delete the selected segments. Click on Submit to confirm. Clicking on Close or on the top right cross will close the window and the segment pairs will not be deleted.

You can delete several segment pairs at once.

Note

Although the confirmation window only displays the source segments, both sides (source and target) will be deleted if you confirm.

Add new segments

To add new segments, click on the + Add new segments button.

A form will be displayed, where you can enter source and target segment pairs. You can add up to five segment pairs at once.

Once you’re done creating your new segments, click on Submit to add them to the corpus. Clicking on Close or on the top right cross will close the window and the segment pairs will not be added.