Tommaso Teofili created OPENNLP-1169:
----------------------------------------
Summary: WordVectorTable should reference WVs by String
Key: OPENNLP-1169
URL: https://issues.apache.org/jira/browse/OPENNLP-1169
Project: OpenNLP
Issue Type: Bug
Components: word vectors
Reporter: Tommaso Teofili
Assignee: Tommaso Teofili
Fix For: 1.8.4
{{WordVectorsTable}} API retrieves {{WordVector}}s via {{CharSequence}}, this
is suboptimal as implementors could store such WVs via an hash table (e.g.
{{MapWordVectorsTable}}) and the value of {{CharSequence#toString}} is not
guaranteed to be the stable.
Additionally it's more common to have words as Strings rather than
CharSequences, being that more consistent with other OpenNLP APIs (e.g.
{{Tokenizer}}).
So {{WordVectorsTable}} should instead retrieve {{WordVector}}s using String.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
