Skip to main content

Short Text Classification

Project description

This project had been deprecated. Maybe use deep learning.

Introduce

A Modification of LibShortText and LIBLINEAR.

  • Uses Wissen Text Analyzer

  • Feature Selection

  • API Exported by Skitai App Engine

  • Win32 support (need MSVC)

Installation

git clone https://gitlab.com/hansroh/haiku
cd haiku
python setup.py build install

Basic Usage

import haiku

model_path = "./golforbed"
analyzer =  haiku.StandardAnalyzer (max_term = 200, stem_level = 2, make_lower_case = 1)

trainset = [
    ('Golf', "cloudy cold calm"),
    ('Golf', "sunny warm"),
    ('Bed', "rainy hot"),
    ('Golf', "sunny hot windy"),
    ('Bed', "windy cloudy cold"),
    ('Bed', "rainy cloudy cold"),
]

# training
h = haiku.Haiku (model_path, haiku.CL_L2, analyzer)
# pruning by document frequency and scoring by meth (FS_CF means category frequency)
h.select (data, mindf = 0, maxdf = 0, top = 0, meth = haiku.FS_CF)
# set training options: uni/bigram and feature representation
h.train (haiku.BIGRAM, haiku.FT_BIN)
h.close ()

# guessing
h = haiku.Haiku (model_path, haiku.CL_L2, analyzer)
h.load ()
print (h.guess ("sunny cold windy"))
h.close ()

Exporting API through Skitai App Engine

Place model data into app_root/resources/haikus/golforbed.

import haiku
import skitai

if __name__ == "__main__":

  pref = skitai.pref ()
  pref.config.resource_dir = skitai.joinpath ('resources')
  skitai.mount ("/", haiku, "app", pref)
  skitai.run (port = 5005)

Go to http://127.0.0.1:5000/haiku/golforbed/guess?q=sunny%20cold%20windy.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

haiku-lst-0.1.1.4.tar.gz (136.2 kB view details)

Uploaded Source

File details

Details for the file haiku-lst-0.1.1.4.tar.gz.

File metadata

  • Download URL: haiku-lst-0.1.1.4.tar.gz
  • Upload date:
  • Size: 136.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: Python-urllib/3.6

File hashes

Hashes for haiku-lst-0.1.1.4.tar.gz
Algorithm Hash digest
SHA256 0a84a43f900a82020e6995556808059e9b797ca97dd688b57c0800b76fb8e20b
MD5 de4a4648aaf2148161e15d1a5d9f5c80
BLAKE2b-256 20ad38f7f7ab630a76f909749565e52ae11421e769b6c8e691d6be6aa07f6a27

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page