How to match the segmentation between Rigi and XTM Cloud?
Problem
The segmentation generated in Rigi and third-party tools can differ. Rigi treats a software string as one segment, even if it contains multiple sentences. If a string containing multiple sentences is imported into XTM Cloud, for example, the results shown below might occur.

A string containing multiple sentences was shown as one segment in Rigi. When an XLIFF file containing the string was exported from Rigi and added to an XTM Cloud translation project, the segmentation in XTM Workbench was different, as shown in this screenshot:

However, the TM in XTM Cloud contained the same (multi-sentence) segment as was present in Rigi.
Cause
The basic issue is that Rigi only manages whole strings (i.e., TUs (translation units)) in an XLIFF, while XTM Cloud can also segment texts at sentence level. This creates a seeming mismatch, but in reality, there is no problem. This is because when a target file is generated in XTM Cloud, the translation units it contains are reassembled at segment level. So, the best solution depends on what is best in a particular case. Some clients working with Rigi use both string- and sentence-level segmentation.
Solution
In XTM Cloud, it is very easy to change the segmentation rules to "paragraph segmentation" mode. Then, entire TU translation units are kept as individual translation segments. This is usually the default behavior for XLF source files in XTM Cloud, so it is important to consider how to change this configuration carefully.
Important
The default behavior for XLF source files in XTM Cloud is that entire TUs are kept together as one translation segment, with no further segmentation of the string that they contain. It is possible, in some systems, that someone deliberately sets the system-level segmentation rules to sentence-level segmentation for some XLIFF files.
XTM Cloud can segment strings at sentence level if no target translations are already present in XLIFF, but that is an additional configuration that must be requested.
Before changing the system-level segmentation rules, we recommend that you investigate where sentence-level segmentation has been specified. Possibilities include a Filter template or an Analysis manager setting. Changing the segmentation rule back to paragraph segmentation might accidentally disrupt translation processes defined by another user.
Possible solution: a new, separate Filter template just for your Rigi file translations. It will enable you to have different segmentation rules from those in other XLF translation workflows.
Issues to consider before changing segmentation configurations
In brief, these are some of the issues to consider before changing segmentation configurations in XTM:
The main disadvantage to using sentence segmentation with Rigi is that, in Rigi previews, translations for multi-sentence strings might be incomplete. In the XTM Workbench screenshot seen above, there is a dotted line between the two segments. This dotted line indicates that the two segments came from the same "block" of text, i.e., paragraph, string, or TU. As mentioned above, this has no negative effect when target translation files are generated, output, and delivered in XTM Cloud. However, in this case, problems might occur in Rigi. When a translation unit contains a multi-sentence string, with each sentence stored in a separate segment, the ‘Rigi signature,’ which tells Rigi which preview to fetch for the translation unit (containing all the individual sentence-level segments), is only stored in the first segment. This prevents translation previews from working perfectly for the second or later sentences in the string.
Paragraph-level segmentation for TUs in XTM Cloud so that they perfectly match the strings in Rigi results in good, clean target previews in Rigi. However, there is some risk of leveraging loss because, for example, repetitive sentences in strings are not matched. This gain in sentence-level leveraging is why some choose to keep sentence segmentation even while using Rigi.
If there is a large body of existing TM for content (created in a long-term project, for example), changing the segmentation configuration in XTM Cloud will cause some leveraging loss. However, it is possible to prepare for this change in configuration and minimize any loss by, for example, using XTM International's Aligner tool to process the current source and target file sets. This migration should be planned together with the XTM Support team. The process should be tested and validated before the big switchover is made. The case described here, moving from sentence to paragraph segmentation, is much easier to manage than the other direction. XTM's Aligner can create the new TM data from the current source & target files, or it might even be possible to dump a bilingual XLF file from Rigi and import it into the XTM Cloud TM to create paragraph-level TM data. However, some data transformation of, for example, inline variables might be needed. The Aligner tool can do this.