We used Elasticsearch in the system. Although I used its analyzers and queries. I did not go deep into its indexing. at the moment, I don’t know how ES allows us to work with the Lucene (inverted) indices that it has in its turtles.
Now we are looking at a series of NLP functions - NER for one thing and Stanford NLP.
There is no plugin to work together these two packages (?)
I did not have a deep look at Stanford NLP. however - as far as I saw, it works all this according to its own indicators. any object or type transferred to it, the Stanford NLP indexes it and gathers from there.
This will make the system work with two different indexes for the same set of documents - those from ES and StanfordNLP, and that would be expensive.
Is there any way around this?
I have one scenario: make StanfordNLP work on Lucene segments - inverted indexes that are already created. In this case:
1.) Does StanfordNLP use Lucene indexes without reindexing anything for themselves? I don't know the StanfordNLP indexing structure - or even how much it uses / doesn't use Lucene.
2.) Are there any restrictions on the use of Lucene indices in ES turtles? would we hit the bottom of the rock using these Lucene segments as is, bypassing the ES between them?
I'm trying to put everything together - everything is in the air. sorry naive Q.
OpenNLP . - , " " ES- (?)
, StanfordNLP.
.