Texty is a collection of simple script frontends to NLP toolkits.
Oftentimes when doing the initial stages of text analysis, you'll want to do some quick and dirty analysis, testing various methods to see what works for your particular application. To help in this, there are many excellent open source NLP tools out in the wild. However, each of these tools has their own pecularities, requirements, data formats etc., and just getting your data into the right format can take quite some time digging through documentation and Java APIs to figure out.
This is where Texty comes in. Texty is essentially a collection of scripts, intended to provide a common command line frontend for NLP tasks: POS tagging, classification, topic modeling, etc. It leverages a number of existing open source tools under the hood for document conversion and algorithm implementations. Essentially, this project is the outcome of my frustration with having to mess with all these tools in a nonuniform manner.
What Texty is not:
- It does not contain any new NLP algorithms. Texty relies on underlying libraries to perform NLP tasks.
- It probably won't give you the state-of-the-art answer. Texty is intended for rapid prototyping purposes.
Run the texty script to see available commands.
Currently integrated:
- Document categorization using OpenNLP:
texty cat,texty cat-train
If you're in Windows, you'll need Cygwin. Otherwise, everything should be bundled with Texty.
- Integrate more functionality
- Test in Linux