An Impose of Dense Neural Network for Detecting Clickbait on Nepali News

Abstract

Purpose: This research aims to detect clickbaits on Nepali news. Clickbaits are frequently existing in online
Nepali digital media. Media house put catchy headlines which, in most of the cases, appears significantly different
from the actual content inside it. They embellish the truth to entice readers to click on it.
Methods: A Machine learning model with Dense Neural Network (DNN) is imposed to train and test on the
Nepali clickbait dataset. The model takes a featured dataset with cosine similarity and Term Frequency Inverse
Document Frequency (TFIDF) to detect clickbaits and non-clickbaits.
Results: Our model achieved a high performance, evidenced by an F1 score of 96.27 on the test data with cross
validation, demonstrating its effectiveness in distinguishing between clickbait and non-clickbait content.
Conclusion: Our study presents a successful application fo dense neural networks for clickbait detection in Nepali
news, offering a valuable tool for improving news consumption quality. Future works will explore expanding the
dataset and incorporating more advanced neural networks.