Categories Busineesses with Text analytics in Python

Question

I'm a new-bee to AI and want to perform the below exercise. Can you please suggest the way to achieve it using python:

Scenario - I have list of businesses of some companies as below like:

 1. AI
 2. Artificial Intelligence
 3. VR
 4. Virtual reality
 5. Mobile application
 6. Desktop softwares

and want to categorize them as below:

 Technology                 ---> Category
 1. AI                      ---> Category Artificial Intelligence
 2. Artificial Intelligence ---> Category Artificial Intelligence
 3. VR                      ---> Category Virtual Reality
 4. Virtual reality         ---> Category Virtual Reality
 5. Mobile application      ---> Category Application
 6. Desktop softwares       ---> Category Application

i.e when I receive a text like AI or Artificial Intelligence, then it must identify AI & Artificial Intelligence as one and the same and put both keywords under Artificial Intelligence category.

The current approach I follow is using the lookup a table but, I want to apply TEXT CLASSIFICATION on the technologies/business for the above input using python where I can segregate the technologies instead of using the lookup table.

Please suggest me any relevant approach.

Kitchenmaid

How much data can you train on? In other words, do you have a lot of data? – Melainemelamed

About 1000records – Kitchenmaid

Answer 1

Here's one approach using sklearn. In past cases, I would use LabelBinarizer() but it won't work in a pipeline because it no-longer accepts X, y as inputs.

If you are a newbie, pipelines can be a bit confusing but essentially they just process the data in steps before passing to a classifier. Here, I am converting X into an ngram "matrix" (a table) of word and character tokens, and then passing that to a classifier.

import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.pipeline import Pipeline, FeatureUnion

X = np.array([['AI'],
       ['Artificial Intelligence'],
       ['VR'],
       ['Virtual Reality'],
       ['Mobile application'],
       ['Desktop softwares']])
y = np.array(['Artificial Intelligence', 'Artificial Intelligence',
       'Virtual Reality', 'Virtual Reality', 'Application', 'Application'])

pipeline = Pipeline(steps=[
    ('union', FeatureUnion([
        ('word_vec', CountVectorizer(binary=True, analyzer='word', ngram_range=(1,2))),
        ('char_vec', CountVectorizer(analyzer='char', ngram_range=(2,5)))
        ])),
    ('lreg', LogisticRegression())
    ])

pipeline.fit(X.ravel(), y)
print(pipeline.predict(['web application', 'web app', 'dog', 'super intelligence']))

Predicts:

['Application' 'Application' 'Virtual Reality' 'Artificial Intelligence']

Melainemelamed

Thanks Jarad. Can you please help me in understanding what have you done.. – Kitchenmaid