Member-only story

Design a personalized news ranking system for Facebook

Pradeep Pujari
6 min readAug 30, 2024

Enhancing User Engagement and Relevance

Few years back I was prepering for Meta on-site interview, there was two 1 hr slot for ML design. I did not find any good material for ML design although there was enough for software engineering system design. Therefore, I put togather a usecase for the benefit of the redears.

Problem Definition:
The problem is that Facebook users are inundated with vast amount of news content on their feeds, leading to information overload and difficulty in finding relevant and trustworthy news articles. Users often encounter content that may not align with their interests or may be biased, leading to dissatisfaction and potential disengagement with the platform. So, there is a need for a system that personalize the news feed for each user, presenting them with articles that are relevant and of high quality thereby enhancing user experience and informed engagement with the platform.

ML Objectives

So, the ML objective is to create an ML model that scores feeds for a specific user, that has high likelyhood of clicking, commenting or liking thus increasing engagement in the platform. So below is the simple diagram of input and output sorted by score.

Scope:of the project underlines what will be included and excluded along with objectives and deliverables.

Inclusions:

  1. To understand individual preferences from analysis of user interactions, preferences, subscription to groups and demographic information
  2. Implementation of personalized news ranking algorithm
  3. Integration of engagement metrics like Likes, shares, comments and reading time to gauge the popularity of news articles.
  4. Testing and refinement of personalized news ranking system based on user feedback and performance metrics.
  5. Ethical considerations including safe guards to prevent the misinformation, hate speech and ensuring diversity and balance in the recommended news articles. (Need to clarify with interviewer)

Exclusions:

In depth content moderation and fact checking.

--

--

Pradeep Pujari
Pradeep Pujari

Written by Pradeep Pujari

AI Researcher, Author, Founder of TensorHealth-NewsLetter, ex-Meta

No responses yet

Write a response