Eric Kafe

In the original WordNet 3.0 database, there used to be endless cycles between 'restrain.v.01' and 'inhibit.v.04' with both hypernyms() and hyponyms(). Currently, nltk uses a patched version of WordNet 3.0,...

In "util.py", "breadth_first" does not call "tree" before looping forever, so changing the definition of "tree" cannot prevent endless cycles in "closure". One possibility is to change the definition of...

Please note however that although @ekaf's proposed solution above for "closure" solves the endless cycles (by checking if a synset was already traversed), it introduces a bug that is worse...

Loops are only endless when they occur in the same path in the tree. Once these are prevented, redundant searches in different pathes are still possible. When the tree is...

The issue with "closure" is different from "tree", because it repeatedly calls an external breadth-first generator, which cannot be interrupted without also loosing any subsequent synset in the relation, and...

Great style, @CubieDev! However, due to the potential severity of this bug, it is preferrable to raise a warning than just to ignore the cycles silently. So, although the fix...

In addition to the small number of endless cycles already found, the theory predicts that closure will also loop endlessly with any call of any symmetric relation, since all symmetric...

@JamesArthurHolland the OMW lines that you quote don't mean that "murciélago" exists as a single word in OMW-1.4, but that it exists as a part of two compounds: >from nltk.corpus...

@fcbond: Of course I will remove these two languages if you insist. My understanding was that these files were ok to put in an omw-extra package, since you still distribute...

According to https://github.com/nltk/nltk_data/pull/171#issuecomment-984387013, "Native speakers of Farsi and Mandarin have pointed out that these two resources have some quality issues". It could be interesting to hear anything about the severity...