hext: a text classification library

[ bsd3, library, natural-language-processing, program ] [ Propose Tags ]

Please see README.md

[Skip to Readme]
Versions [faq],,,,
Dependencies base (>=4.7 && <5), containers, hext, text, unordered-containers [details]
License BSD-3-Clause
Copyright 2016 David Anekstein
Author David Anekstein
Maintainer aneksteind@gmail.com
Category Natural Language Processing
Home page https://github.com/aneksteind/hext#readme
Source repo head: git clone https://github.com/aneksteind/hext
Uploaded by aneksteind at 2016-09-17T15:06:40Z
Distributions NixOS:
Executables hext-exe
Downloads 2380 total (20 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Hackage Matrix CI
Docs available [build log]
Last success reported on 2016-11-20 [all 1 reports]




Maintainer's Corner

For package maintainers and hackage trustees

Readme for hext-

[back to package description]


This is currently the beginning of a text classification library.

##Installation/Running stack install hext

hackage - https://hackage.haskell.org/package/hext-

To run:
stack build
stack exec hext-exe


Currently, the only algorithm implementation is the Naive Bayes algorithm: to run your own data through this algorithm in order to classify your text, you need:

  • classified data: this can be sourced from a database where the only fields that are needed are the text itself, and it's class
  • a sample string which will be classified by the algorithm

In order to run the program, the classified data specified above must be converted into a BayesModel a using the teach function, where a is your own defined data type representing the class to classify your text. Your class must be and instance of Ord and Eq.

With your new BayesModel filled with knowledge, it's time to classify your text using runBayes. An example of this can be seen in app/Main.hs where data Class = Positive | Negative deriving (Eq, Ord, Show) to label movie reviews as either positive or negative.


I encourage contributing to this package, in the form of implementing algorithms that are not yet in the project, improving efficiency, or the like.