dkpro / dkpro-core-examples

Ready-to-use examples of dkpro-core components and pipelines.
Other
34 stars 17 forks source link

Full NLP4J example - lexica file not found #28

Closed ggiavelli closed 5 years ago

ggiavelli commented 5 years ago

Was having trouble getting a full NLP4J example to work. I want to segment to sentences and tokens, run POS, Lemmatization, Morphological, and Dependency.

I will update this with the specific error I'm getting with looking for a file "lexica..."

reckart commented 5 years ago

I don't find an NLP4J example in this repository. Where did you find the example that is not working?

ggiavelli commented 5 years ago

I figured it out. It was a dependency thats not in maven central. No there isnt a lot of java examples so no full NLP4j example. I'm on a different computer now or I would share it

I am doing a eval of different modules so maybe I can post some examples for java

I was wondering.. .I am trying to convert a dependency tree into an object from the output So I'm starting with the Dependency object but... its line by line it's a bit tricky to convert it to a graphical object representation. Any ideas?

On Sun, Mar 3, 2019 at 5:11 AM Richard Eckart de Castilho < notifications@github.com> wrote:

I don't find an NLP4J example in this repository. Where did you find the example that is not working?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/dkpro/dkpro-core-examples/issues/28#issuecomment-469012409, or mute the thread https://github.com/notifications/unsubscribe-auth/AFiM29XaPIK1FIeFwzvZDyO-Nmh_kj5xks5vS63SgaJpZM4bVz3T .

--

Gianna Giavelli skype: Gia.Giavelli

reckart commented 5 years ago

You might care to have a look at the BratWriter- it has a mode of rendering annotations as HTML. Here yo can see some screenshots taken during the development process: https://github.com/dkpro/dkpro-core/issues/656

reckart commented 5 years ago

Btw. if you aren't aware if it yet: you could probably save your data as XMI or WebAnno TSV and import it into an annotation editor like INCEpTION (or WebAnno).

ggiavelli commented 5 years ago

Thanks Richard.

I think i'm not so interested in display, as for my SOLR indexing project I am trying to generate tuples (quadruples) which map to terms and their dependency relationships.

e.g. DO:Microsoft-VP:Acquired VP:Acquired-PO:Documentum

Probably I will just create my own parse of the dependency object. But if you are familiar with anything I might tryit. I'll def look at WebAnno format and see if I get a nicer object

Cheers!

Dr. Giavelli

On Sun, Mar 3, 2019 at 4:00 PM Richard Eckart de Castilho < notifications@github.com> wrote:

Btw. if you aren't aware if it yet: you could probably save your data as XMI or WebAnno TSV and import it into an annotation editor like INCEpTION https://inception-project.github.io (or WebAnno https://webanno.github.io/webanno/).

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/dkpro/dkpro-core-examples/issues/28#issuecomment-469069152, or mute the thread https://github.com/notifications/unsubscribe-auth/AFiM26SRBIC1IHsRDOC1n6MAqHiqLrJ3ks5vTEXugaJpZM4bVz3T .

--

Gianna Giavelli skype: Gia.Giavelli

reckart commented 5 years ago

First exporting the data in some format and then reading it again from there to convert it to a third format seems to be an unnecessary complication. I think you'd be better off implementing a custom UIMA writer component in Java which accesses the data directly in the UIMA CAS object and then send it off to your Solr index - maybe using our SolrWriter component as a starting point.

reckart commented 5 years ago

I gather the general problem has been resolved, so I'm closing this issue.