Possibility to write pipeline components in Python

zesch / pydkpro

5 stars 1 forks source link

Possibility to write pipeline components in Python #11

Open zesch opened 4 years ago

zesch commented 4 years ago

It should be possible to also have Python-based components and mix them with DKPro processing.

aggarwalpiush commented 4 years ago

I am not sure about Python-based components. It would be great to have some examples. Please acknowledge, if your intention of mixing components like NumPy arrays or pandas data frames with DKPro processing.

zesch commented 4 years ago

I was more thinking of writing a UIMA component interfacing the Cassis CAS directly in Python.

reckart commented 4 years ago

I can think of two scenarios:

a Python-based analysis engine, e.g in order to call the NLTK tokenizer or POS tagger. The analysis engine would need to pick up all relevant information from the CAS, do its processing and then write updates back to the CAS
a Python-based writer, e.g. to draw a diagram or something similar. The writer would pick up data from the CAS and write the results to screen or to a file.