It seems that the prefix matching of a phrase examines the first word and then next, separately.
First, I t thought "terminal" was caught by dint of stem lookup, but when it's turned off, the result is the same.
I'd appreciate it if some improvement can be made in this area.
I am not entirely sure what you mean but in your example the prefix matching works as expected:
'terminal' word contains a prefix - 'term'
'terminal service' phrase contains a prefix - 'term service'
The prefix matching is meant to increase the likelihood of catching a word or a phrase in the current segment. You can control it with the prefix length option.
> when both prefix matching and stem lookup are enabled, which takes preference? Are both results shown?
First, it stems the word and if the stem is long (longer than the prefix length option), it shortens it to the prefix.
>In this example, "terminal service" (glossary phrase) would be recognized (depending on the % setting) as a match for source segment phrases like "terminal S" or "terminal serv" (if any!), and not for >"term service."
It actually looks for phrases starting with "term serv" as in many languages the infection may take place in each word of the phrase.
>> It actually looks for phrases starting with "term serv" as in many languages the infection may take place in each word of the phrase.
Today, CafeTran presented me "failure to comply" (glossary term) as a match for "fails to comply" (source segment).
On a second thought, this is more effective in that it can catch more.