Challenge background
Vosoughi, et al. mentioned that in Twitter from 2006 to 2017 about 126,000 rumors were spread by ∼3 million people, where the top 1% of false news cascades diffused to between 1000 and 100,000 people, whereas the truth rarely diffused to more than 1000 people. Such relevance of not being informed adequately, motivated at the early beginning of this year (2022) to the LATAM Chapters of Omdena (Mexico, San Salvador and Colombia) to promote and build together their first MVP of fake News Validator in spanish with manual labeling of news and tweets.
Project timeline
- 1
Week 1
Understand the problem, and make brainstorming.
- 2
Week 2
Link search engine APIS. Definition of heuristics for the context engine searcher.
- 3
Week 3
Webscrapping. Definition of heuristics for the context engine searcher.
- 4
Week 4
Semantic Modeling. Topic Modeling. Research of context engine models. Overview of BLOOM from Hugging Face. Overview of GPT 3 from Open AI
- 5
Week 5
Semantic Modeling. Topic Modeling. Research of context engine models. (Only if feasible ). Apply BLOOM from Hugging Face (Only if feasible ). Apply GPT 3 from Open AI
- 6
Week 6
Semantic Modeling. Topic Modeling. Research of context engine models. (Only if feasible ). Apply BLOOM from Hugging Face (Only if feasible ). Apply GPT 3 from Open AI
- 7
Week 7
Research of context engine models. (Only if feasible ). Apply BLOOM from Hugging Face (Only if feasible ). Apply GPT 3 from Open AI. Streamlit App (Optional)
- 8
Week 8
Unify and merge the most relevant models respectively. Streamlit App (Optional). Present Results
What you'll learn
Learn how to make connections with APIs of searching engines. Learn how to web scrape news and tweets. Learn how to do a topic modeling analysis. Learn what algorithms / models exist in the state of the art for the search or comparison of the context. Learn how a context engine works.