shandilya21 / news-headline-classification-based-on-genres

Deep Learning technique for news headlines classification

News Headlines Classfication based on Genres

This model which describes the genre based news headlines supervised clustering techniques for easy accessible of information This Model designed for word embedding from Deep Contextual representation of Language Model. Here we introduced a model that will generate a word embedding by looking into the context.Here we take 'News Headline' and its 'Category' along with its short description. Here we need to classify the Headline based on its Category. There are total 22 catedory like "Entertainment","Politics","Sports" etc. Along with that Short description for formatting the headline and defining the headline properly. wwe used the Kaggle news dataset for performing this task and try to incorporate the model a state of teh art result.

Models

• Model_1 : CNN - CNN Model
• Model_2 : CNN -LSTM Model

Result:

Genres includes Entertainmen, sports, movies, politics, business etc slong with other 22 classes with the 10000, raw text. we simulated our result with the following metrics: Precison, recall, F1-measure

• Model_1 : Precison: 0.78, Recall: 0.77, F1-score : 0.783
• Model_2 : Precison : 0.83, Recall:0.81, F1_score : 0.846

Requirements and packages/Library:

• Python 3.5+
• Tensorflow
• Keras
• Numpy
• Pandas

.