BUG: Match all word forms problem in French
The problem was originally brought to attention here (French thread): https://community.logos.com/forums/p/127665/831054.aspx#831054
After investigation, I found that match all word forms (abbr. MAWF) will, in the NASB95 produce horse, horses, etc when I search for "horse". However, in the LSG, with MAWF on, the search for "enfant" (child) does not produce the many instances of "enfants" (children) and vice-versa. I only have one other French resource (Massorah Gedolah: Manuscrit B. 19a de Léningrad) and it does the same thing (try "liste" vs. "listes").
I am not sure whether this applies to French as a language or to some French resources only, or whether the problem is also found in other languages including some English resources.
Comments
-
The MAWF algorithm is designed to work with English word stems; not olde English nor any other language.
A search for "enfant" with MAWF will produce "enfants", but a search for "enfant, enfants" without MAWF produces more results. It appears that MAWF ignores "enfants" when tagged as French language, but not when searching for "enfants"! You can see this clearly with a normal search for "french:enfants"
Dave
===Windows 11 & Android 13
0 -
Thanks, Dave, for the explanation. Although it may have been designed that way originally, the software has since expanded considerably toward the use of other languages. It is a liability that searches do not work optimally in them and since the fonctionality is available to be checked on/off in other language interfaces as well, it should be made to work properly in them as well.
I am not sure whether you tested the search for "enfant" with MAWF on in a different resource than mine, but mine (in Bible Louis Segond) does not produce "enfants" at all:
I hope that FL will look into this.
0 -
Francis said:
I am not sure whether you tested the search for "enfant" with MAWF on in a different resource than mine, but mine (in Bible Louis Segond) does not produce "enfants" at all:
As I said above "It appears that MAWF ignores "enfants" when tagged as French language, but not when searching for "enfants"! So if it is fully tagged as French there will be no result! What happens if you search for "enfants" with MAWF.
My library has no French resources but some have text tagged as French.
Dave
===Windows 11 & Android 13
0 -
Francis said:
I am not sure whether this applies to French as a language or to some French resources only, or whether the problem is also found in other languages including some English resources.
Stemming (the technology behind Match All Word Forms) is not currently implemented for French.
0 -
Thanks, Bradley, for chipping in.
Presently, I don't use French resources much anyway. However, speaking on behalf of French-speaking users (see the cross-referenced thread above), it would seem to me that ability to perform such basic searches reliably across languages should be a must. It seems problematic to me, from a "sellability" standpoint, if it starts to show that there are a number of functionalities that might not work properly in other languages because they were originally designed for use with English resources (especially if it is not clear upfront what these limitations might be). In other words, these need to be remedied and (if it is not done already) my suggestion would be for a team to identify other language-dependent features that might need corrective work now that Logos is reaching out to so many other markets internationally.
0 -
Improving foreign-language support is high on our priority list.
This typically requires us to bring a native speaker on staff, so that they can test that the program performs as expected, provide missing translations, and help English-only software developers who have questions about how the software should perform.
0
