cankayauniversity/ceng-407-408-2018-2019-multi-class-classification-of-news-text-1-

Multi-class Classification of News Text (1)

ceng-407-408-Multi-class-Classification-of-News-Text-1-

Introduction

Nowadays, there are a lot of informations on Internet and every information has their own classification which can be related with medical, marketing, news and many more. Every information has and needs some kind of classification and it needs to be classified for quick access and preventing the data loss. The classification has been widely studied and it has more than one way to classify an information with computer which can be generalizable by one main topic which is Machine-learning.

Purpose

Purpose of this document is explain how to install and use the Multilabel News Text Classification as Web Application.

Prerequisites / Tools

This web application can work in local or in an Virtual Private Server (VPS) (i.e. AWS Lightsail). However it may not work in Cpanel web hosting (regular website hosting) since building this project needs CLI for installing Python and PHP packages/libraries.

If you are planning to use it on local you can use Xampp for Windows and Mamp for MacOS.

1. Classification

The text classification needs Magpie library in the backend. Which is written in Python language. So install Python from here as 64-bit since Tensorflow dependency can work properly only on 64-bit. If you install 32-bit it may not work. Magpie library needs Tensorflow, Gensim, nltk, keras.

Use

pip install <<package_name>>

for installing packages.

2. Web Application

For executing ".py" (Python Script File Extension) in web application change the file below.

Open "\httpd.conf" file in Apache directory. And add:

AddHandler cgi-script .py

ScriptInterpreterSource Registry-Strict

to the end of file.

ceng-407-408-2018-2019-multi-class-classification-of-news-text-1-'s Issues

Classification with a different dataset using Magpie

Koral try a different dataset with magpie. Write here the results. Accuracy of the model and more. Here is a list of rare datasets:

https://medium.com/@dataturks/rare-text-classification-open-datasets-9d340c8c508e

Dataset

Need text datasets so we can test on Magpie.

After founding datasets the files should be this format

Separate label and text file.

Web app

• Login