Wiktionary lookup bug about vocabsieve HOT 5 CLOSED

Hi,
Thanks for testing the tool!

This is actually the intended behavior. When lemmatize is set to true, it will follow the first link to the original form of the word, which will result in it omitting all other entries. In this case, it went to "livrer" because it finds a span.form_of tag in the definition.

The intended way to avoid this is the other button, "Define (direct)" that appears when you set "lemmatize" to true, which will temporarily set it to false.

from vocabsieve.

In this case, that doesn't quite make sense - since livre is the lemma for the noun forms of the word as well, which have an entirely different meaning than that of the verb form returned by the function. So it is not the case that the function is returning "the original form of the word" for livre in the sense of "book."

from vocabsieve.

Well, unfortunately the library used to do the lemmatization is a simple database of words (think of a giant dict) and can't really handle homonyms reliably. From the wiktionary() function there's also no way to tell which sense is the intended one.

Though, the use of following links in wiktionary function is a historical thing, now that we have a dedicated library for that purpose. I might as well remove it because of these unintended effects. Btw, does this still happen if you choose another dictionary option (they don't follow links).

from vocabsieve.

Tbh I have only played with this Python function, not the GUI yet : )

Another option might be adjusting the implementation of the wikipedia lookup so that it doesn't exit too early. That is, if it finds a word is a form_of another word, rather than returning the recursive call to itself on the link it finds for the lemma word, it could just add the result to a list of definitions being collected. It may need to maintain a set of lemmas already looked up in order to avoid duplicates in the output list, though.

from vocabsieve.

The commit 7060465 should remove the recursive lookup behavior. In addition while investigating this I found another important bug in the fmt_result function which causes the output to be repetitions of the first entry rather than display all entries in a sequence. Thanks for this issue! Now the wiktionary function will no longer act differently depending on lemmatize option, since that task has been delegated to the preprocessing part.

from vocabsieve.