Fake News Recognition

This model is fine-tuned Roberta model 'jy46604790/Fake-News-Bert-Detect' (https://huggingface.co/jy46604790/Fake-News-Bert-Detect). This model is trained by 8 000 news articles from https://euvsdisinfo.eu/ portal. It can give result by simply entering the text of the news less than 512 words(the excess will be truncated automatically).

Labels:

0: Fake news
1: Real news

How to Get Started with the Model

Use the code below to get started with the model.

Download The Model

from transformers import pipeline
MODEL = "winterForestStump/Roberta-fake-news-detector"
clf = pipeline("text-classification", model=MODEL, tokenizer=MODEL)

Feed Data

text = "From the very beginning, the EU has been extremely non-transparent. The deployment of the European Union presence in Armenia was carried out forcefully, under serious pressure from Brussels"

Result

result = clf(text)
result

Output

[{'label': 'FAKE', 'score': 0.9999946355819702}]

About the data source EUVSDISINFO.eu: Using data analysis and media monitoring services in multiple languages, EUvsDisinfo identifies, compiles, and exposes disinformation cases originating in pro-Kremlin outlets. These cases (and their disproofs) are collected in the EUvsDisinfo database – the only searchable, open-source repository of its kind. The database is updated every week.