I built a concordance for the Roman Martyrology and let it show me connecting words while grouping similar forms. It grouped "for" with "Forli", which is a place in Italy with wet winters, warm summers, and apparently at least one martyr or two. It's hard for me to verify that last part because even with Forli in quotation marks in Inline Search, Verbum [Logos] still won't let me distinguish between place and preposition and gives me 818 or less false positives.
Please fix this. Thanks!
Please use descriptive thread titles to attract helpful posts & not waste others' time. Thanks!
SineNomine: It's hard for me to verify that last part because even with Forli in quotation marks in Inline Search, Verbum [Logos] still won't let me distinguish between place and preposition and gives me 818 or less false positives.
Turn off "Match all word forms" in your Search tab. It affects inline search as well, even though it doesn't seem connected. I was able to find the one instance of Forli in that resource using inline search.
But you're right, when "Match all word forms" is on, searches for "for" and "Forli" should not be finding each other.
How to Ask for Help | Logos Wiki | My Machine Specs | My Blog
This is a "known issue" with stemming (match all word forms / group similar forms); see https://community.logos.com/forums/p/87708/615401.aspx#615401
I can see why there would be some errors in stemming due to using an algorithmic stemmer, but when is "li" an English ending that should be stripped off by a stemmer? Is it confusing "li" for "ly"?
Rosie Perera: I can see why there would be some errors in stemming due to using an algorithmic stemmer, but when is "li" an English ending that should be stripped off by a stemmer? Is it confusing "li" for "ly"?
My guess is that since stemmers have support for removing chains of inflection, "li" handles cases like: "uglier" -> "ugli" -> "ugly"
Here's some more info: http://snowball.tartarus.org/algorithms/english/stemmer.html
Andrew Batishko | Faithlife software developer
Is it possible for Faithlife to manually override the imperfect algorithm?
SineNomine: Is it possible for Faithlife to manually override the imperfect algorithm?
Yes, but see caveats discussed here: https://community.logos.com/forums/p/61695/437957.aspx#437957
Bradley Grainger (Faithlife):Yes, but see caveats discussed here: https://community.logos.com/forums/p/61695/437957.aspx#437957
Am I understanding correctly the the caveat is that it's a pain in the neck to do?
SineNomine: Bradley Grainger (Faithlife):Yes, but see caveats discussed here: https://community.logos.com/forums/p/61695/437957.aspx#437957 Am I understanding correctly the the caveat is that it's a pain in the neck to do?
I think the caveat is that it might break something, because there are complicated dependencies.
Rosie Perera:I think the caveat is that it might break something, because there are complicated dependencies.
If a specific exception of the type "'Forli' is not a form of the word 'for'" breaks something (I guess one might also need to define what other forms of Forli there might be-I'm thinking "Forli's" and nothing else), then that would seem to indicate that there's something else more seriously wrong.
Incidentally, an end user tool for making precisely this kind of separation, though quite possibly impractical or nearly impossible to program, would make me quite satisfied.
Since this ability/inability seems like it would be directly connected to "fuzzy search," perhaps Faithlife should simply change the last phrase in this screenshot!
Diagnostic Logs