StanfordNLP NE Scorer node doesn't work

Hello everybody,

the StanfordNLP NE Scorer node doesn't work for me. It shows the following warning: "Tokenizer "null" could not be found, due to missing language extension! Install additional language extensions at File->Install KNIME Extensions."

I am new to KNIME and thus have the latest version including its text processing extensions. Nevertheless, I followed these instructions and installed every extension that is related to language. Morevover, the software is updated on an on-going basis.

I asked a friend of mine who is also new to KNIME and she has the same problem.

Which language extension is required in particular?

It would be very nice if someone could help me out.

 

Thank you in advance and best regards,

Lauritz

Hi Lauritz,

Could you please share an example workflow that shows this error message?

Just as a reminder, you need to use StanfordNLP NE Learner node before the StanfordNLP NE Scorer. The StanfordNLP NE Learner node creates a conditional random field model based on documents and a dictionary with entities that occur in the documents. The chosen tag can be used by the StanfordNLP NE tagger node to tag new documents and validate the model.

The StanfordNLP NE Scorer node calculates some quality measures based on the model applied on the data being analyzed.

Hope that helps,

Best,

Vincenzo

 

Hi Vincenzo,

thank you for your answer. Please find attached a screenshot of the relevant part of the workflow.

 

Best regards,

Lauritz

Hi Lauritz,

are you using the same tokenizer in the scorer node as in the learner node? You can check this in the dialog of the nodes.

To check if the nodes have been installed / updated properly please download this example workflow from the Example server and execute it.

knime://EXAMPLES/08_Other_Analytics_Types/01_Text_Processing/14_NER_Tagger_Model_Training

This workflow should work. If not it might be that the installation of the Textprocessing extension or the update was not successful.

Cheers, Kilian

Hi Kilian,

thanks for your answer. I cannot see an option to choose a tokenizer in the scorer node dialog. However, regarding the learner and predictor nodes the tokenizers are equivalent.

Unfortunately, I cannot open the workflow that you have attached. The connection to the server failed.

 

Best regards,

Lauritz

Hi Lauritz,

There might be some issues with the connection with the EXAMPLES Server. Could you please try again to connect to it? It should work just fine.

Thank you,

Cheers,

Vincenzo

Hi Vincenzo,

the problem is that I cannot connect to the EXAMPLES Server in general. The login always fails.

 

Best regards,

Lauritz

After using the zip folder relating to the link above, the Scorer Node displays the same warning.

So far, the text processing plugin has worked very well. I cannot explain why exactly this one thing doesn't work.