Spacy dependency parser

[fvillarroel]

Hello!

I wanted to ask if there is a way to use dependecy parser for text within KNIME.

I know spacy can help in doing that DependencyParser · spaCy API Documentation. So might be there is a way to do it using the spacy extension within KNIME. Currently we don’t have a node, but might be with some python integration after using spaCy pos tagger?

This link shows an example of the dependency parser (tree) of spaCy ML Reference | Dependency trees with spaCy

Thanks!

[steffen_KNIME]

Dear Francisco @fvillarroel,

maybe that would interesting for @abderrahimala from @Redfield, who currently maintains that extension. But maybe @nemad could also assess whether that is even an option within that extension.

Best regards
Steffen

PS: @fvillarroel your link in your profile is broken

[fvillarroel]

Thank you @steffen_KNIME, let’s see what they say.

I updated the link

[fvillarroel]

Hi @Redfield @abderrahimala @nemad do you have any suggestion on this? Thanks

[abderrahimala]

Hi @fvillarroel

Sorry for the delayed response.

As you mentioned, currently we don’t have any node to implement the Dependency Parser. Adding it to the spaCy extension would be a valuable enhancement for future updates .

I am not sure how you are planning to use the Python script node after the POS tagger node. Probably you will need to access the spaCy model inside the Python script node, and unfortunately, you can’t use the model you loaded before using the spaCy model selector. You will need to load it again using the spaCy library, and I am not sure if you will be able to access the document returned by the POS tagger node inside the Python script node either.

[fvillarroel]

Thank you for the response @abderrahimala.

I believe this is the types of things that could really make a difference in using spaCy in KNIME, because it addresses one limitation of KNIME native pre-processing nodes. This is going o become a recurrent need.

Cheers!