Terminology recognition in OmegaT is handled via tokenizers. Starting with version 3.0.0, tokenizers are included in the standard OmegaT distribution, whereas one had to download them separately in previous versions. They are also automatically selected during the project creation process, whereas one had to launch them via the command line in previous versions. Tokenizers are especially important for terminology recognition in heavily inflected languages. This video shows how the tokenizer works with Finnish as the source language.
Starting with OmegaT version 3.0.1, recognized terminology can be inserted in the target segment via a new auto-completer feature, which works entirely in the editor pane and with the keyboard (the shortcut is Ctrl+space in Windows, and Esc in OS X, so as to stay consistent with the system-wide completion engine). In previous versions, one had to right-click with the mouse in the glossary pane. This video shows how terminology can be inserted in the target segment, using a sample Finnish-English project.
Finnish tokenizer (Lucene)