Social media

Social media recommendation systems: The key to unlocking user engagement
Ahsan Manzoor
| January 2, 2023

Billions of users use various social media daily and see a lot of new suggestions there. The content includes text, images, videos, and so on depending on the social platform. Do you know how that content is suggested? 

We will learn about it in this blog.
 

Recommendation system: 

It is an algorithm that suggests relevant products to users based on a variety of factors. Sometimes, when you search for a certain product on a website you notice that you start receiving several suggestions of similar products, there is a system behind this. It is generally used to target potential users more efficiently and improve the user experience by suggesting new items, saving users’ time, and narrowing down the set of choices. 

 

Learn about Data Science here

 

Watch the video to see what a recommendation system is and how it is used in various real-world applications. 

 

Introduction to Recommender Systems 

 

 

Now that we know the concept, let’s dive deeper into a real-world application to better comprehend it. 

 

YouTube’s recommendation system journey

YouTube has over 800 million videos, which is about 17,810 years of continuous video watching. It is hard for a user to repeatedly search for certain sorts of videos from millions of videos. This problem is solved by recommendation systems, which provide relevant videos based on what you are currently watching.

The system also works when you open YouTube’s home page and do not watch any videos. In this case, it shows the mixture of the subscribed, most up-to-date, promoted, and most recently watched videos.  

Let’s discuss the journey of the recommendation system on YouTube. 

In 2008, YouTube’s recommendation system ranked videos based on popularity. The issue with this approach was sometimes violent or racy videos get popular. To avoid this, YouTube built classifiers to identify this type of content and avoid recommending them. After a couple of years, YouTube started to incorporate video watch time in its recommendation system.

The reason for this was that users often watched different types of videos and there were different recommendations for them. Later, YouTube took surveys where users rated the watched videos and answered the questions upon giving low or high stars.  

Soon, YouTube’s management realized that everyone did not fill out the survey. So, YouTube trained a machine learning model on completed surveys and predicted the survey responses. YouTube did not stop there; they started to consider the likes/dislikes and share information to make the recommender system better.  

Nowadays, they are also using classifiers to identify authoritative and borderline (doesn’t quite violate community) content to make a better recommender system. 

 

Read more about social media algorithms in this blog

 

Before diving deep into the technical detail, let’s first discuss common types of recommendation systems. 

Classification of recommendation system:  

Recommendation system
Recommendation system

 

These types of recommendation systems are widely used in industry to solve different problems. We will go through these briefly. 

 

1. Content-based recommendation system

 According to the user’s past behavior or explicit feedback, content-based filtering uses item features (such as keywords, categories, etc.) to suggest additional items that are similar to what they already enjoy. 

Content based recommendation system
Content based recommendation system

 

 

2. Collaborative recommendation system 

Collaborative filtering gives information based on interactions and data acquired by the system from other users. It is divided into two types: memory-based, and model-based systems. 

 

a) Memory-based system 

This mechanism is further classified as user-based and item-based filtering. In the user-based approach, recommendations are made based on the user’s preferences that are similar to the preferences of other users. In the item-based approach, recommendations are made based on items similar to other items the active user likes. 

 

Let’s see the below illustration to understand the difference:  

User-based recommendation system
User-based and item-based recommendation system

  

b) Model-based system 

This mechanism provides recommendations by developing machine learning models from users’ ratings. A few commonly used machine learning models are clustering-based, matrix factorization-based, and deep learning models.  

Model-based system
Model-based system

2. Demographic-based recommendation system 

This system provides recommendations based on user demographic attributes, such as age, sex, and location. This system uses demographic information, such as a user’s age, gender, and location, to provide personalized recommendations. This type of system uses data about a user’s characteristics to suggest items that may be of particular interest to them.

For example, a recommendation system might use a user’s age and location to suggest events or activities in the user’s area that might be of interest to someone in their age group.

 

 

3. Knowledge-based recommendation system 

This system offers recommendations based on queries made by the user rather than a user’s rating history. Shortly, it is based on explicit knowledge of the item variety, user preference and suggestion criteria. This strategy is suited for complex domains where products are not acquired frequently, such as houses and automobiles. 

 

4. Community-based recommendation system 

This system provides recommendations based on user-interacted items within a community that shares a common interest. A community-based recommendation system is a tool that uses the interactions and preferences of a group of people with a shared interest to provide personalized recommendations to individual users.

This type of system takes into account the collective experiences and opinions of the community in order to provide personalized recommendations.

 

5. Hybrid recommendation system 

This system is a combination of two or more discussed recommendation systems such as content-based, collaborative-based, and so on. Sometimes a single recommendation system cannot solve an issue, thus we must combine two or more recommendation systems. 

We now have a high-level understanding of the various recommendation systems. Recall the YouTube discussion, what do you think, which recommendation method suits YouTube the most. 

 

It is a memory-based collaborative recommendation system. YouTube can use an item-based approach to suggest videos based on other similar videos using users’ ratings (clicked on and watched videos). To determine the most similar match, we can use matrix factorization. This is a class of collaborative recommendation systems to find the relationship between items’ and users’ entities. However, this approach has numerous limitations, such as  

  • Not being suitable for complex relations in the users and items 
  • Always recommend popular items 
  • Cold start problem (cannot anticipate items and users that we have never encountered in training data) 
  • Can only use limited information (only user IDs and item IDs)  

To address the shortcomings of the matrix factorization method, deep neural networks are designed and used by YouTube. Deep learning is based on artificial neural networks, which enable computers to comprehend and make decisions in the same way that the human brain does.

Let’s watch the video below to gain a better understanding of deep learning.

 

 

YouTube uses the deep learning model for its video recommendation system. They provide users’ watch history and context to the deep neural network. The network then learns from the provided data and uses the softmax classifier (used for multiclass classification) to differentiate among the videos. This model provides hundreds of videos from a pool of over 800 million videos. This procedure was named “candidate generation” by YouTube.  

But we just need to reveal a few of them to a certain user. So, YouTube created a ranking system in which they provide a rank (score) to each of a few hundred videos. They used the same deep learning model that assigns a score to each video for this. The score may be based on the video that the user watched from any channel and/or the most recently watched video topic.  

User history and context
User history and context – Source 

Summary 

We studied different recommendation systems that can be used to address various real-world challenges. These systems help to connect people with resources and information that may not have been easily discoverable otherwise, making them a useful tool for solving these challenges.

We discussed the journey of YouTube’s recommendation system, a collaborative system used by YouTube, and examined how YouTube performed well using deep learning in their systems.  

Ayesha Saleem
| September 13, 2022

In today’s blog, we will try to understand the working behind social media algorithms and focus on the top 6 social media platforms. Algorithms are a part of machine learning which has also become a key area to measure success of digital marketing; these are written by coders to learn human actions. It specifies the behavior of data by using a mathematical set of rules 

According to the latest data for 2022, users worldwide spend 147 minutes, on average every day on social media. The use of social media is booming with every passing day. We get hooked up on the content of our interest. But you cannot deny that it is often surprising to experience the content we just discussed with our friends or family.  

Social Media algorithms

Social media algorithms sort posts on a user’s feed based on their interest rather than the publishing time. Every content creator desires to get the maximum impressions on their social media postings or their marketing campaigns. That’s where the need to develop quality content comes in. Social media users only experience the content that the algorithms figure out to be most relevant for them.  

1. Insights into Facebook algorithm 

Facebook

Facebook had 2.934 billion monthly active users in July 2022.  

Anna Stepanov, Head of Facebook App Integrity said “News Feed uses personalized ranking, which considers thousands of unique signals to understand what’s most meaningful to you. Our aim isn’t to keep you scrolling on Facebook for hours on end, but to give you an enjoyable experience that you want to return to.” 

On Facebook, which means that the average reach for an organic post is down over 5 percent while the engagement rate is just 0.25 percent which drops to 0.08 percent if you have over 100k followers. 

Facebook’s algorithm is not static, it has evolved over the years with the objective to keep its users engaged with the platform. In 2022, Facebook adopted the idea of showing stories to users instead of news, like before. So, what we see on Facebook is no longer a newsfeed but “feed” only. 

Further, it works mainly on 3 ranking signals: 

  • Interactivity:

The more you interact with the posts from one of your friends or family members, Facebook is going to show you their activities relatively more on your feed.  

  • Interest:

If you like content about cars or automobiles, there’s a high chance Facebook algorithm will push relevant posts to your feed. This happens because we search, like, interact or spend most of our time seeing the content we like.  

  • Impressions:

Viral or popular content becomes a part of everyone’s Facebook. That’s because the Facebook algorithm promotes content that is in general liked by its users. So, you’re also more likely to see what’s everyone talking about today.  

2. How does YouTube algorithm work 

Youtube

There are 2.1 billion monthly active YouTube users worldwide. When you open YouTube, you see multiple streaming options. YouTube says that in 2022, homepages and suggested videos are usually the top sources of traffic for most channels. 

The broad selection is narrowed on the user homepage on the basis of two main types of ranking signals.  

  • Performance:

When a video is uploaded on YouTube, the algorithm evaluates it on the basis of a few key metrics: 

  • Click-through rate 
  • Average view duration 
  • Average percentage viewed 
  • Likes and dislikes 
  • Viewer surveys 

If a video gains good viewership and engagement by the regular followers of the channel, then the YouTube algorithm will offer that video to more users on YouTube.  

  • Personalization:

The second-ranking signal for YouTube is personalization. In case you love watching DIY videos, YouTube algorithm processes to keep you hooked on the platform by suggesting interesting DIY videos to you.  

Personalization works based on a user’s watch history or the channels you subscribed to lately. It tracks your past behavior and figures out your most preferred streaming options.  

Lastly, you must not forget that YouTube acts as a search engine too. So, what you type in the search bar plays a major role in shortlisting the top videos for you.  

3. Instagram algorithm explained  

Instagram

In July 2022, Instagram reached 1.440 billion users around the world according to the global advertising audience reach numbers.  

The main content on Instagram revolves around posts, stories, and reels. Instagram CEO Adam Mosseri said, “We want to make the most of your time, and we believe that using technology [the Instagram algorithm] to personalize your experience is the best way to do that.” 

Let’s shed some light to the Instagram’s top 3 ranking factors for year 2022: 

  • Interactivity:

Every account holder or influencer on Instagram runs after followers. Because that’s the core to getting your content viewed by the users. To get something on our Instagram feed we need to follow other accounts. As much as our interaction with someone’s content occurs, we will be able to see more of their postings.  

  • Interest:

This ranking factor has more influence on reels feed and explore page. The more you show interest in watching a specific type of content and tap on it, the more of that category will be shown to you. And it’s not essential to follow someone to see their postings on reels and explore the page. 

  • Information:

How relevant is the content uploaded on Instagram? This highlights the value of content posted by anyone. If people are talking about it, engaging with it, and sharing it on their stories, you are also going to see it on your feed. 

4. Guide to Pinterest algorithm 

Pinterest

Being the 15th most active social media platform, Pinterest had 433 million monthly active users in July 2022.  

Pinterest is popular amongst audiences who are more likely interested in home décor, aesthetics, food, and style inspirations. This platform carries a slightly different purpose of use than the above-mentioned social media platforms. Therefore, the algorithm works with distinct ranking factors for Pinterest.  

Pinterest algorithm promotes pins having: 

  • High-quality images and visually appealing designs  
  • Proper use of keywords in the pin descriptions so that pins come up in search results. 
  • Increased activity on Pinterest and engagement with other users. 

Needless to mention, the algorithm weighs more for the pins that are similar to a user’s past pins and search activities. 

5. Working process behind LinkedIn algorithm  

LinkedIn

There are 849.6 million users with LinkedIn in July 2022. LinkedIn is a platform for professionals. People use it to build their social networks and have the right connections that can help them succeed in their careers.  

To maintain the authenticity and relevance of connections for professionals, the LinkedIn algorithm processes billions of posts per day to keep the platform valuable for its users. LinkedIn’s ranking factors are mainly these: 

  • Spam:

LinkedIn considers post as spam if it contains a lot of links, has multiple grammatical errors, and consists of bad vocabulary. Also, avoid using hashtags like #comment, #like, or #follow can flag the system, too. 

  • Low-quality posts:

There are billions of posts uploaded on LinkedIn every day. The algorithm works to filter out the best for users to engage with. Low-quality posts are not spam but they lack value as compared to other posts. It is evaluated based on the engagement a post receives. 

  • High-quality content:

You wonder what’s the criteria to create high-quality posts on LinkedIn? Here are some tips to remember: 

Easy to read posts 

Encourages responses with a question 

Uses three or fewer hashtags 

Incorporates strong keywords 

Tag responsive people to the post 

Moreover, LinkedIn appreciates consistency in posts, so it’s recommended to keep your followers engaged not only with informative posts but also conversing with users in the comments section.  

6. A sneak peek at the TikTok algorithm 

TikTok

TikTok will have 750 million monthly users worldwide in 2022. In the past couple of years, this social media platform has gained popularity for all the right reasons. The TikTok algorithm is considered as a recommendation system for its users.  

We have found one great explanation of TikTok “For You” page algorithm by the platform itself: 

“A stream of videos curated to your interests, making it easy to find content and creators you love … powered by a recommendation system that delivers content to each user that is likely to be of interest to that particular user.” 

Key ranking factors for the TikTok algorithm are: 

  • User interactions:

This factor is like the Instagram algorithm, but mainly concerns the following actions of users: 

Which accounts do you follow 

Comments you’ve posted 

Videos you’ve reported as inappropriate 

Longer videos you watch all the way to the end (aka video completion rate) 

Content you create on your own account 

Creators you’ve hidden 

Videos you’ve liked or shared on the app 

Videos you’ve added to your favorites 

Videos you’ve marked as “Not Interested” 

Interests you’ve expressed by interacting with organic content and ads 

  • Video information: 

Videos with missing information, incorrect captions, titles, and tags are buried under hundreds of videos being uploaded on TikTok every minute. On the discover tab, your video information signals tend to seek for: 

Captions 

Sounds 

Hashtags* 

Effects 

Trending topics

  • TikTok account settings:

TikTok algorithm optimizes the audience for your video based on the options you selected while creating your account. Some of the device and account settings that decide audience for your videos are: 

Language preference 

Country setting (you may be more likely to see content from people in your own country) 

Type of mobile device 

Categories of interest you selected as a new user 

Social media algorithms relation with content quality 

Apart from all the key ranking factors for each platform, we discussed in this blog, one thing remains ascertain for all i.e., maintain content quality. Every social media platform is algorithm bsed which means it only filters out the best quality content for visitors. 

No matter which platform you focus on growing your business or your social network, it highly relies on the meaningful content you provide your connections.  

If we missed your favorite social media platform, don’t worry, let us know in the comments and we will share its algorithm in the next blog.  

Related Topics

Top
Statistics
Programming Language
Podcasts
Machine Learning
High-Tech
Events and Conferences
DSD Insights
Discussions
Development and Operations
Demos
Data Visualization
Data Security
Data Science
Data Engineering
Data Analytics
Computer Vision
Career
Books
Blogs

Up for a Weekly Dose of Data Science?

Subscribe to our weekly newsletter & stay up-to-date with current data science news, blogs, and resources.