Detecting interested topics on social media
The increasing amount of information on social media and news sites has made it challenging for readers to filter out irrelevant news. To address this issue, deep learning and NLP technology have been used to build text classifiers to categorize the latest news. This paper aims to develop an e...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/167193 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | The increasing amount of information on social media and news sites has made it
challenging for readers to filter out irrelevant news. To address this issue, deep learning
and NLP technology have been used to build text classifiers to categorize the latest
news. This paper aims to develop an end-to-end system that can deliver different
interesting topics of news to users. The system includes a news extraction API, news
dataset collection, a model that can classify the latest news, and a UI interface to show
different topic news to users.
This paper compares the performance of different classification models, including TF IDF connected to Naive Bayes, LightGBM, KNearestNeighbours, Random Forest,
XGBRF, and embedding layer developed with Tensorflow Keras connected to RNN,
CNN, and LSTM. The best models was then applied to real-world data to filter daily
latest news articles for users. The study updates the training dataset to keep up with
vocabulary trends, ensuring better news categorization.
Finally, a Streamlit framework was used to deploy the model into a web application
with a good user experience. This project report provides a comprehensive guide for
building a news classification system using deep learning models and deploying them
on a web application. |
|---|