Closed jmccrae closed 1 month ago
This ignores the possible presence of a head in a sense key. If lemmas can have unescaped %, so do heads, leading to possible unparsability of sense key. This does not happen with the current data set and is unlikely. But unlikely things happen sometimes.
Lemmas with '%' are potentially ambiguous as discussed in #1123 as this leads to two percentage (
%
) occurring in the sense key.This PR fixes our tools to work with them as follows.
The
lemma
and thelex_sense
are split by the last percentage sign to occur. In this way ambiguity is avoided.This even works with the Princeton WordNet tools: