How to train social media algorithms so that only interesting content is in your feed
Miscellaneous / / April 03, 2023
How social media algorithms work
Smart recommendations are the result of the work of developers and machine learning specialists. They write a program code with algorithms that collects information about user actions and remembers personal preferences. Also, artificial intelligence can analyze the behavior of your friends and subscribers on the social network, the popularity of content among the entire audience, geolocation and other indicators.
Algorithms study not only the format of posts and their authors, but also the subject matter. Frequently liking recipes on social networks means they will appear in the recommendation feed. If favorites change over time, the system notices this and rebuilds. As a result, a feed filled with potentially interesting posts is formed.
The operation of the recommendation system depends on the conditions set by the creators of the algorithms. Here are the possible scheme her behaviour:
- Collaborative. Collects information about user actions and looks for similar behavior patterns among other accounts. Then it compares preferences and gives advice based on likes and subscriptions of people with similar interests.
- Content. Focuses solely on the actions of the user himself - what topics and formats he liked, such AI will recommend.
- hybrid. Combines the first two methods in different proportions and combinations. For example, may prefer content or "friends of interest", or may consider them equivalent.
- Subject. Provides recommendations for predefined filters. Most often used in online stores.
In the new Russian social network ЯRUS Several algorithms are responsible for the selection of recommendations at once. And they are simultaneously available to all users. So, the most popular news and topics of the day fall into the category "Trends" - likes and comments are not taken into account when forming it. In the “News by Interests” section, the user himself chooses what topic he is interested in, and the social network displays a selection of relevant posts. In Geography News ЯRUS analyzes geolocation and suggests publications related to a specific city or region.
Finally, "Personal Tiers" is a feed that fully meets the user's preferences. There you can specify interesting topics, favorite formats and content sources. For example, form a tier from text posts about a story from other users or from fashion videos from media blogs. You don’t need to limit yourself to one selection: the social network allows you to create several personal tiers. Such feeds provide an opportunity to meet users with similar interests and unite to create collaborations.
Customize your tiers
How to teach artificial intelligence to show what you are interested in
Regardless of which scheme the recommender system uses, its cool offers are the result of constant training. Both developers and users can train algorithms. Here's how to do it.
1. Always authenticate
Many social networks, streaming services and other entertainment platforms allow you to consume content or some part of it without registration. But you won’t be able to set up a personal recommendation feed in this way: you will only see posts that are popular now or relevant to your location. Authorization will allow the platform to accumulate knowledge about you and your preferences and improve the recommender system.
2. Interact with content
Like, bookmark, leave comments - show the algorithms that you liked what you saw. And vice versa, if the social network has the option to "Hide posts" of a certain user, public or channel, mark publications that you are not interested in. Over time, the algorithm will learn your preferences - there will be practically no boring content in the feed.
3. Declutter
If your profile has a lot of pages that you no longer look at, the tips will suffer. Therefore, it is important to periodically clean up the subscriptions.
Some social networks can do some of the work for you: for example, evaluate recent profile interactions and suggest whose content you usually scroll through without likes and comments. Another way to help is to distribute subscriptions according to the frequency of interactions: in such a list, the most favorite pages will be at the top, and long-forgotten ones at the bottom.
4. Respond to system recommendations
Another way to improve AI is to check its "homework". Go to the page with recommended content and give feedback in the form of clicks on posts and likes. When the system recommends something wrong, report the error using the "Not interested" option. It hides in a pop-up menu - depending on the social network, you can open it by clicking on the three dots next to the publication or by long pressing on the post. If you often evaluate the recommendations of the system, it will quickly turn into an excellent student.
5. Enter more information about yourself
The recommendation system can be guided not only by your reactions, but also by demographic indicators. To make the advice in the feed more accurate, fill out the profile completely: indicate gender, age, place of residence.
Another beacon for the system is geolocation. If you want to see posts that are relevant in your region, allow the application to receive location data, if not, disable it.
IN ЯRUS find interesting posts easily. For this, in addition to smart recommendations, there are keywords and hashtags. The first ones are automatically assigned to each post - a smart system is responsible for the distribution. Hashtags are added by the authors of the publications themselves. To find a suitable post, the user can either indicate their interests or type in the desired key in the search bar.
AI TI SERVIS LLC
Price: Free
Download
Price: Free
AO Dialog
Price: Free
Download
Price: Free
Also in ЯRUS there is an event feed with cool events nearby, video and music. The last section will allow you to discover new artists - tracks are added to the site by independent artists. Making money with a blog is also real: the social network has tools to support content with coins - the internal currency of ЯRUS. And the competition among the authors here is still small - it's time to declare yourself and become a paid content creator.
Start a blog in YARUS