Problem Statement

According to a recent report from Mozilla, although facing many urging for crucial reforms, YouTube’s algorithm is still really problematic when fueling extreme opinions of hate and violence.

“71% of all Regret reports came from videos recommended to our volunteers and recommended videos were 40% more likely to be regretted than videos searched for.” - Mozilla

Additionally, YouTube updated its terms of service last year, and it says that YouTube will run ads on smaller creators' videos without paying them. The change is very unfair to the smaller content creators out there because to earn money through YouTube, they will then have to become a YouTube Partner with very demanding criteria.

“You grant to YouTube the right to monetize your Content on the Service (...). This Agreement does not entitle you to any payments.” - YouTube

We want to solve these two problems by creating a more inclusive YouTube recommendation algorithm and home page that detects and removes videos that might express extreme opinions and disinformation and boosts smaller content creators.

Product Specifics

The final product should meet these requirements documented below.

Algorithm specifics

• Fast detection and removal of videos that might contain extreme opinions/beliefs or misleading information
  • using machine learning to determine a pattern based on user input/feedback on random videos they watch/got recommended by the algorithm
  • a more prominent ‘Report’ button (replacing the ‘Clip’ button) on the video page
  • analyzing comments on the video to prevent content creators from abusing the algorithm
• Promotion of more quality content created by small content creators on the recommendation wall

Home page specifics

• On the home page, categorize the recommended videos sections such as “Recently watched,” “New/rising content creators,” etc

Built With

• data
• fastapi
• python
• react
• youtube