Ultimate Guide to Decoding the Influential Web Algorithms

1
Ultimate Guide to Decoding the Influential Web
Algorithms
www.hexwhale.com
2

• Earlier in the 20th century, life was simpler and
  harder at the same time. We werent aware of what
  else we might want to purchase on Amazon. We were
  not aware of the most important news/stories of
  the day. Before the invent of a movie
  recommendation engine, we had no mechanized
  assistance to determine which DVD to buy next.
  Google helps us to think, while Facebook helps us
  find friends. So, the web algorithms that these
  services embed are actually easing our problems
  and making our lives stress free.
• Also, web algorithms do have a huge impact on our
  unique tastes, buying habits, and decisions
  regarding our digital lives. When were looking
  for something online, Googles algorithm is right
  there to solve our issue of sorting and searching
  through a number of not very relevant results. On
  the contrary, these web algorithms might trap us
  in a world where brands and businesses couple
  behavioral data with website formulas to predict
  and manipulate what we decide next.

3

• The technical trend towards the more
  sophisticated web algorithms isnt confined to
  situations where consumers seek information about
  products/services. Businesses are harnessing the
  power of these web algorithms to boost their
  efficiency in dealing with inventory control and
  their efficiency in monitoring behavior. They can
  also predict what their customers next move
  might be.
• Interestingly, the Internet is a Candyland of
  data to model predict behavior with. Tracking
  IP addresses across the Net, analyzing what kind
  of websites people visit and when they visit
  these websites, counting on clicks all are much
  easier with the implementation of web algorithms.
• Here is a look at some of the amazing web
  algorithms and the interesting ways in which they
  work

4
1. YouTube Up Next

• These are a personalized list of videos that a
  viewer may have an interest in watching next,
  depending on their previous YouTube activities.
  These videos appear on the side panel of the
  YouTube interface or as Watch Next if Autoplay
  is on. One user will not have the same experience
  as the other on YouTube. However, YouTube aims to
  serve the most relevant, personalized
  recommendations to each of its viewers. To
  achieve this, YouTube first analyzes users
  viewing history and identifies hundreds of videos
  that might be relevant to them.
• Then, these videos are ranked depending on how
  well each video engages users, how often each
  viewer watches videos from each channel or videos
  relevant to the same topic, and the number of
  times YouTube has already shown each video to
  users.
• Also, users tend to watch more content when they
  receive recommendations from a variety of
  channels.

5
2. Facebook News Feed

• The algorithm behind the Facebook news feed was a
  rank-based approach called the Edge Rank
  Algorithm. This algorithm determines the order of
  posts based on three parameters Affinity Score,
  Edge Weight Time Decay.
• Currently, the news feed algorithm has undergone
  a revamp through a machine learning approach. The
  algorithm considers more than 10000 weights and
  focuses on posts that are predicted to promote
  active engagement. The term active engagement
  means that the algorithm predicts scores by
  assigning greater weight to parameters that make
  a post personal worthy of conversation.
• The four parts of this algorithm are as follows

6

• Inventory
• This comprises of all lined up posts that are yet
  to be viewed by the end user. These posts include
  promotional content, posts from pages followed,
  etc. A number of such posts must compete with one
  or the other post each day to be visible to the
  algorithmic arbiter. In the end, only a few of
  these make it up to the news feed of the user
  after they pass through the algorithms filter.
• Signals
• Signals are all about the content consideration.
  Each post is analyzed based on the available data
  such as
• Number of likes, comments shares
• Type of Post (video, images, textual)
• Time and Day of a post
• Blocked Content

7

• Time spent on post
• Top fifty interactions
• Video engagement
• Predictions
• The algorithm attempts to predict based on
  information available to determine what the users
  prefer to see on their feed, what they may hide,
  how probable are they to engage with it actively.
  If video content is seen to be receiving higher
  engagement over text content or pictures, such
  posts are predicted to be preferred by the user.
• Scoring
• The predicted posts along with the weights are
  used to arrive at a relevancy score. The posts
  are then ordered based on this score in
  descending order. These posts are then delivered
  in the determined sequence to the news feed.

8
3. Google Search Rankings

• With the amount of information available on the
  web, finding what we require would be nearly
  impossible without some sorting through it.
  Google ranking algorithms are updated to do the
  sorting through billions of webpages in our
  search index. This is done to find the most
  relevant, useful information in a fraction of a
  second and present them before us helping us find
  what we were looking for.
• These ranking systems are made up of a whole
  series of algorithms. Search algorithms consider
  many factors such as the words of our query,
  relevance, and usability of web pages, the
  expertise of sources, our location, and settings.
  The weight applied to each criterion varies
  depending on the nature of the query.
• To ensure search algorithms satisfy high
  standards of relevance and quality, a rigorous
  process is in place that involves both live tests
  thousands of trained external Search Quality
  Raters from around the world. These Quality
  Raters abide by strict guidelines that define the
  goals for Search algorithms and are available for
  anyone to see.

9
4. Instagram Feed

• The Instagram algorithm helps to surface the best
  and the most relevant content to each user
  every-time they check their Instagram feed. The
  whole point of this algorithm is to serve you the
  posts that really matter to you.
• Instagram offers posts, using Machine Learning,
  that it thinks users would like and interact
  with. It is dependent on the users historical
  usage the type of posts they liked and
  interacted with in the past as well as on the
  owner of the post. Also, the six main factors
  that determine what you see in your Instagram
  feed are
• Interest
• Timeliness
• Relationship
• Frequency
• Following
• Usage

10
5. Youtube Related

• Before Youtube generates video recommendations
  for viewers, the system first determines a set of
  relevant videos that a user is likely to watch
  after viewing a seed video. For this, it uses a
  method called co-visitation/association rule
  mining to identify pairs of videos watched during
  a particular session. Then, a relativity score is
  calculated for these videos. The relevant videos
  association rules are then combined with a users
  activity on the site, such as videos watched and
  liked by the user, to create a seed set. Once
  this is done, the paths of related videos can be
  traced out from this seed set to generate
  recommendations.
• Once a set of video recommendations have been
  generated, they are ranked according to certain
  signals, which can be put into three groups
• video quality
• user-specific, and
• diversification

11
6. Instagram Explore

• Instagram has been working hard to make the
  Explore page more personalized to each individual
  user.
• When we view our Instagram Explore page, it looks
  different than what may show up on others
  explore page. Instagram Explore is a gateway to
  discover content. And, this gateway is
  specifically tailor-made to help us find new
  content relevant to what we would enjoy watching
  on Instagram.
• The Instagram Explore algorithm is constantly
  observing and analyzing from individual human
  behaviors of what accounts you follow, what posts
  you like and comment on what Instagram stories
  you love to watch. It then uses this data to
  create an Instagram Explore page with suggestions
  of similar content that you might like.

12
7. Twitter Feed

• The primary domain of Twitters feed algorithm is
  the Top Tweets. Top Tweets are visible to you
  depending on accounts that you interact with,
  tweets that you engage with, and so on. The
  essential ranking signals used by the Twitter
  feed algorithm are
• Recency
• How recent the tweet is.
• Engagement
• No. of retweets, clicks, favorites, and
  impressions a tweet received.
• A tweets audience engagement is relative to
  other tweets from the same account.
• How often people interact with the tweets
  author, through active engagements.
• Rich Media
• The type of content present in a tweet (image,
  video, GIF) and even polls.
• The type of media users tends to engage with.

13
8. Reddit Popular

• Reddit always uses a story algorithm called as
  the Reddit hot ranking. With the Reddit hot
  ranking algorithm, the number of votes and link
  publishing time have the largest effect on the
  rank of a story.
• Reddit implements a logarithm function in its
  algorithm. Thus, the first votes on a link are
  more valuable than the later votes on a link. As
  a link gets older, its ranking also will slowly
  plummet. However, it is also essential to get
  some initial traction on a submission to give it
  early visibility.
• Reddit ranks a post by calculating the number of
  votes a link has and then subtracting points
  based on how old that link is. So, recent links
  generally rank higher than older links. This
  updates the front page as well as ensures that
  links with highest up-votes arent on the front
  page for weeks or months together.

14
9. Instagram Suggested Follows

• The Instagram Suggested Follows algorithm is
  pretty basic because the suggestions are based on
  whom you have searched before. It implies that
  whose account shows up at the top of your
  suggestions list is someone you would have
  searched for a number of times. The Instagram
  suggestion algorithm isnt solely based on how
  often you look for a specific account. Even after
  we delete search suggestions, it will use some
  similarities that it found before you deleted the
  Instagram search history to make those accounts
  reappear in other ways.

15
10. Google Discover

• A summary of your page can be presented as a card
  that will be shown to users in Google Discover.
  This would be a scrollable list of topics that
  users can browse on their mobile devices. Tapping
  a card will direct the users to a page that is
  the source of the Discover entry.
• Discover displays a mix of content to users based
  on their interactions with content/products on
  Google that they choose to follow directly.
  Google Discover is not confined to what is
  published today. Rather, if a user finds the
  earlier content interesting, then Discover will
  show it.
• Discover also features videos, sports,
  entertainment updates, stock prices, and much
  more. Google Discover is a hub of content
  pertaining to the users interests.

16
THANK YOU