OmegaT now offers a GUI alignment tool, accessible via Tools > Align Files. The tool supports all file formats supported by OmegaT. The workflow has two parts:
- Automated alignment:
- Text is extracted from the files, segmented, and then given an initial automatic alignment according to algorithms provided by the mALIGNa library.
- Aligned sentences are grouped together in 1–n relationships, indicated by the row highlighting in the table. A set of contiguous cells sharing the same color will end up squashed to a 1–1 text unit in the final output.
- The user may tweak algorithm parameters, segmentation rules, file filter settings, etc. to improve the automatic alignment before proceeding.
- Manual adjustment:
- The user can manually edit the aligned sentences. Moving up/down, splitting, merging, and editing are supported.
- Groups can be included or excluded from the final output via the "Keep" checkboxes. Groups where source and target are identical are excluded by default.
- Groups can be marked as "accepted" or "needs review".
- Efficiently leverage the alignment algorithm and minimize manual shuffling with Pinpoint Align and Realign Pending:
- Identify misaligned regions.
- Mark the region's borders (the previous or next correctly aligned groups) as "accepted", or use
Edit > Start/End Pinpoint Align
(Space
key) to create borders.
- Automatically realign the not-yet-accepted regions with
Edit > Realign Pending
(Cmd/Ctrl+R
).
When manual adjustment is complete, the user can save the result as TMX.
Other features:
- Customizable regex-based text highlighting to help visually assess alignment. Numbers (
\d+
) are highlighted by default.
- All editing commands are accessible from the keyboard.
- Emacs-like table navigation keys.
- ID- or path-based alignment for file formats that support it.
- "Parsewise" alignment for files that happen to contain the same number of pre-segmentation entries (e.g. files translated with OmegaT).
- Optionally remove all tags from the text.
The Pinpoint Align and Realign Pending features are available in r8639.
Implemented in the released version 4.0 of OmegaT.
Didier
Samuel wrote a tutorial:
http://www.proz.com/forum/omegat_support/306343-new_interactive_aligner_in_omegat.html
Didier