连接到专用服务器-CoreNLP

Question

I am looking for the python code required to connect to a local instantiation of the Stanford CoreNLP Server.

I have successfully connected and communicated with Stanford's server where you need to enter the following into the command line prompt:

java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000

and have nlp = StanfordCoreNLP('http://localhost:9000') in your python code.

I now have a local CoreNLP server operating on CentOS 6. I use the terminal prompt: sudo service corenlp start and get the system response CoreNLP server started. Which I assume means my local instantiation of the CoreNLP server is listening for any requests. I now need to know how to communicate with my local server.

Can anyone enlighten me on what I need to replace nlp = StanfordCoreNLP('http://localhost:9000') with in order to be able to talk to my local CoreNLP Server?

Answer 1

You should use the Stanford NLP group's Python library stanza :

https://github.com/stanfordnlp/stanza

Hopefully over the next month there will be better documentation and support for starting and stopping the server in Python.

Answer 2

As noted above by @StanfordNLPHelp, stanza can be imported and used with the following code:

from stanza.nlp.corenlp import CoreNLPClient
client = CoreNLPClient(server='http://localhost:9000', default_annotators=['ssplit', 'tokenize', 'lemma', 'pos', 'ner'])
annotated = client.annotate('This is an example document. Here is a second sentence')

This code acts in a similar manner to that of this line of code, which should look familiar if you have been connecting to Stanford's server nlp = StanfordCoreNLP('http://localhost:9000') .

My Problem was that I could not connect to the dedicated server that resides on my local CentOS based operating system.

The Solution to my problem was that I was missing the 'middleman'. By that I mean I was missing the part of the process where I connected to my dedicated server. I had the dedicated server properly set up and running, and I had the Python code that would reach out to the server BUT I did NOT have the "pipeline" open to my dedicated server.

To do this I just needed to follow these simple steps:

1. Download CoreNLP 3.7.0 to Downloads folder.

   a. unzip folder. Note which location you extracted to

2. Disconnect from the Internet!!

3. Enter the following into the terminal window:

   a. cd ~/stanford-corenlp-full-2016-10-31 (<--wherever you stored this file)

   b. java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 80 -timeout 15000

4. Run Python code.

*Note that I disconnected from the internet. If you do not do this the java command will not connect to your local server but Stanford's.

Summary: When using Stanford CoreNLP there are three components that you need (1) Appropriate line of code in Python script to call out to the server. Two options presented above (2) A functioning server. The two options I am aware of include Stanford's own server, or a dedicate server you have on your local system. and (3) the CoreNLp software that opens the pipeline from your system to the server. Number three is where I was getting caught up.

*I am interested to know if there is a way to connect to my local server without disconnecting the internet. It seems that if the internet is enabled the code will default to Stanford's instanciation of the server. Is it possible to overwrite this default? Feel free to comment below with input on this side bar.

EDIT: I do not believe that the above information is correct. According to Stanford CoreNLP's Server page , "Stanford CoreNLP ships with a built-in server, which requires only the CoreNLP dependencies." I believe that my solution above was only using this "built-in" server not my local dedicated server. I am still investigating this issue. I will edit as more information is uncovered.