Twitter Design

  • Business Function?
  • Registration?
  • Security?
  • Scalability?

** Further define the problem, make it clear and more concrete
** Focus on Core Features

Business Function

  1. Follow People
  2. View Other Feeds

Data Modelling


Feed Serve - Get all feed from people you follow and render them by time (order)

Q1. So many feeds, how many can they be rendered on 1 view?
  • Infinite scroll, only N feeds are fetched at a time.
  • Backend cache
Q2. Detect Fake Users - Machine Learning
  • Registration Date - Valid?? Very young vs very old
  • No. of Followers - Extraordinary Large
  • No of Feeds - Extraordinary Large

Tweet Ordering using some methods

  • User Liking
  • User relationship to the author. Number of replies/retweets to his tweets
  • No. of followers of the author
  • User favourite number of tweets by some author 

Trending Topics

  • Most frequent hashtags over last N hours
  • Hottest search queries
  • Most favourite tweets over last N hours and extract common words or phrases

Ranking

  • Frequency (retweets may be) (Topic - # or query or words/phrases)
  • Favourite/ Reply/ Retweet
  • Follows/Followers talking about that topic

Who to Follow

  • May be friends of friends

        P Follows R. R is the potential person you may know

  • Celebrities

       People with many number of followers e.g celebrities

Ranking

  • User has previously followed this user
  • No. of common follows/followers
  • Overlap in basic information (e.g. location)

Moments

Interesting Topics of the moment (new, sports, fun etc.). Tweets related to these categories.

New articles for the last 1-2 hours. Associate topic with each article. Find tweet with each topic.

Trending Topics - Categorization
Rank Topics
Rank tweets in topics
Content, Comments find popular
Categories - sports, fun, food etc.

Search

  • Search on tweets
  • Build Indexing, ranking and retrieval
  • Ranking Freshness/social signals
  • Associate weights on feature/signals Features - reply, retweet, favorite numbers, users popularity, freshness. Add up and display according to weight.



Comments

Post a Comment

Popular posts from this blog

Terraform Basics

Image
Disadvantages of Data Center Deployment Time to set up Many manual steps involved Under utilization of resources e.g. unused web sites during night time Costly IaC (Infrastructure as Code) Hence IaC comes into picture which is API based and provisions resources in cloud  environment. Basically it can configures the entire provisioning environment. Types of infrastructure components are web servers, databases, networks 3 types of Iac 1. Configuration Management Ansible Puppet SaltStack      on existing Infrastructure componenets. 2. Server Templating Docker Packer Vagrant Usually create images on VM or containers 3. Provisioning Tools Terraform Cloud Formation Provision Infrastructure components using simple declarative code. Terraform - Infrastructure Provisioning tool for various cloud platforms. Providers Providers enable Terraform to provision various infrastructure components. Infrastructure Components on cloud platforms Network Components - BigIp, Infoblox Data ...
Read more