Collabs/Veracity_AI
Collabs
/
Veracity_AI
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
69 Commits 3 Branches 0 Tags 2.1 MiB
Python 100%
Go to file
Falko Victor Habel e960ba886b
Gitea Actions Demo / Explore-Gitea-Actions (push) Successful in 9m53s Details
Merge pull request 'develop' (#18) from develop into main 
Reviewed-on: Berufsschule/Veracity_AI#18
 		2024-12-05 11:16:03 +00:00
.gitea/workflows .gitea/workflows/run.yaml aktualisiert 2024-11-01 06:59:07 +00:00
Token added really good token checking 2024-10-16 09:26:14 +02:00
docs updated code 2024-12-05 09:05:17 +01:00
src updated code 2024-12-05 09:05:17 +01:00
tests updated pytests for new langchain libary 2024-12-05 09:41:04 +01:00
theme added docs 2024-09-05 11:36:08 +02:00
.gitignore added really good token checking 2024-10-16 09:26:14 +02:00
LICENSE added docs 2024-09-05 11:36:08 +02:00
README.md updated code 2024-12-05 09:05:17 +01:00
fake_news_checker.db fix database 2024-10-10 09:29:27 +02:00
requirements.txt added tests and final code cleaning 2024-10-30 19:09:25 +01:00

README.md

Veracity_AI

Overview

Veracity_AIn is designed to analyze articles and determine their authenticity using Natural Language Processing (NLP) techniques and machine learning models. The application consists of two main components: a user interface for inputting URLs or text, and a backend system that processes the data and makes predictions about the veracity of the article.

User Interface

Fake News Checker UI

The user interface is built using CustomTkinter, a modern and customizable Python UI library. The main screen is divided into three sections:

  1. Input Section: This section allows users to enter the URL of an article or paste the text directly into the input box.
  2. Result Section: Displays the result of the analysis (Real or Fake) and the confidence level in percentage.
  3. Leaderboard Section: Shows a list of news providers along with their fake news percentages, sorted by the highest fake news rate.

Backend System

The backend system is responsible for processing user input, communicating with the database, and making predictions using the VeraMind model. Here's an overview of its components:

Data Models

  • TextData: Stores the URL, text content, provider, result, confidence, and a flag indicating if the news is fake.
  • Provider: Represents a news provider with attributes for name, total articles, fake articles count, and a list of associated TextData objects.

Database

The application uses a DuckDB database to store analyzed data. The FakeNewsChecker class manages database operations such as inserting new data and fetching existing data.

Machine Learning Model

  • VeraMindInference: An inference engine for the VeraMind model, which is used to predict whether an article is real or fake news based on its text content.
  • ArticleRater: A class that uses the Large Language Model (LLM) to generate a response based on the analyzed text data.

Usage

To use the Fake News Checker application, follow these steps:

  1. Enter an article URL or paste the text directly into the input box.
  2. Click on the "Check" button to initiate the analysis process.
  3. Once the analysis is complete, the result (Real or Fake) and confidence level will be displayed in the result section.
  4. The leaderboard section will automatically update with the latest news providers' fake news percentages.

License

This application is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International. See the LICENSE file for more details.