My Projects

This project was developed to analyze the real-time development of news coverage through a dashboard that gathers data from various news publishers via Google News. It utilizes a Multi-Agent System powered by ChatGPT-4.0-mini to perform sentiment analysis, pre-processing, and topic categorization. The dashboard enables users to access information quickly and efficiently, providing valuable insights into current phenomena or issues related to specific topics.

Login Page View

In this login page, the system is also equipped with security features that ensure user credentials are transmitted to the database in an encrypted format, maintaining the confidentiality and integrity of login information.

News Article Scraping Page

In this project, I used Selenium and cookie management to automate the news article crawling process. The bot is designed to bypass bot detection mechanisms on certain news websites, ensuring smooth and uninterrupted data collection. The scraped data is automatically stored in the database, with a built-in system to prevent duplicate entries.

Dashboard Visualization

The dashboard is equipped with various visualization features to support in-depth analysis, including an article timeline that displays publication distribution over time, a general categorization chart to show overall topic classification, and separate charts for positive and negative news coverage to identify sentiment trends. It also features a list of the top 15 publishers along with their sentiment tendencies, as well as a Textual Network Analysis that visualizes the relationships between topics or keywords in the form of a text-based network.

Raw Data Overview

On this page, the project focuses on the processing and cleaning of raw data collected from various sources. Steps such as handling missing values, data normalization, and feature transformation are carried out to prepare the data for further analysis. The page displays the collected raw data and includes a customizable feature that allows users to update sentiment and manually categorize topics. This functionality is especially useful for handling cases where the automated bot may have misclassified sentiment or topics, ensuring more accurate and reliable analytical results.