websocket-tagger

==========

Description

This project provides a WebSocket server that wraps the Stanford Part-of-Speech tagger. This makes it easier to get part-of-speech tags from JavaScript for arbitrary text. Please note that this project is just a prototype that illustrates the utility of WebSockets for NLP.

Requirements

• Java
• Ant
• Stanford CoreNLP
• Java WebSocket
• A modern browser that supports WebSockets (See http://caniuse.com/websockets). I recommend the latest version of Mozilla Firefox.

Installation:

1. Clone the Java WebSocket repository and build java_websocket.jar (you may also download the jar directly from the repository). Add this jar to your CLASSPATH.
2. Download stanford-corenlp-1.3.5.jar and stanford-corenlp-1.3.5-models.jar and add them to your CLASSPATH (Note: make sure to download the latest version which at this point happens to be 1.3.5)
3. Run ant to build the StanfordTaggerServer class.
4. Run the included script start_server.sh to start the WebSocket server.
5. Load tag.html into Firefox.
6. Tag away.

Notes:

1. Please note that currently the server needs to be killed manually since this is what I want. Obviously, this can be easily changed with the same thinking that goes into how to properly kill a regular socket server.
2. It should be simple to extend the server to do anything else CoreNLP can do besides tagging.

Screenshot:

Here's a screenshot of the jQuery powered tag.html page in action: