spaCy is the best option to handle the dependency parser with Python. You can see more here https://spacy.io/usage/visualizers

Actually, NLTK contains Stanford Dependency Parser and Stanford Parser and Stanford Tagger. Could you describe your problem exacter?

Hello Dr. Osherenko ,

Thank you verymuch for your reply. I have tried using sanford dependency parser from nltk. Unfortunately the library returns only the best tree as an output but it doesnot give information of how many different trees are possible for a given sentence.

I would like to know the alternative trees possible for a given sentence.

Thanks and Regards

Udaysimha Nerella

Hello Mr. Vitor Rolim,

Yes spacy is a good library. But I didnot find the solution what I am looking for using spacy.

Thanks and Regards

Udaysimha nerella

Hi Udaysimha Nerella,

I would like to know the alternative trees possible for a given sentence.

I doubt you can extract such information using the Stanford Parser since the Stanford authors make only such information accessible that is necessary for the users of the parser. Information regarding the number of possible trees is too specific for the Stanford implementation. You can, however, try other options:

1. You can patch the Stanford parser and extract information you need. Python code (stanford.py) is only an interface to the Java engine that starts a Java process in the _execute function. Download the Java source code of the parser and debug it.

2. If you don't want to debug, it is probably easier to compare the parsing trees from different parsers. As others say, Stanford, spacy etc.

3. It is not necessary to stick to parsers. Maybe, you could use taggers for your analysis, for example, the Stanford tagger and the Stanford parser (both in the NLTK as Python interfaces to Java engines). Taggers are rather quicker and maybe sufficient for your analysis.

Best, Alexander

Hi Udaysimha,

Did you find an answer to your question?