| Subject: | Bug Report |
| Date: | Mon, 15 Jul 2019 12:44:15 -0400 |
| To: | <bug-Word2vec-Interface [...] rt.cpan.org> |
| From: | "Scott Guthery" <sbg [...] acw.com> |
Perl: V5.26.1
Platform: Ubuntu 18.02.2 LTS
Word2vec::Interface: 0.031
I am using ReadTrainedVectorDataFromFile, GetVocabularyHash, and ComputeCosineSimilarity to get a similarity measures between all word pairs in the data file.
1) Two words are added to the keys of the vocabulary hash returned by GetVocabularyHahs: the number of unique words in the file (404 in my case) and the string ‘<\s>’
2) In some small number of cases, ComputeCosineSimilarity yields a string value of the similarity measure rather than a numerical value.
Many thanks for your contribution to CPAN.
Cheers, Scott