A recommendation system is an algorithm or tool that predicts and suggests items, content, or services to users based on their behaviors, preferences, and interactions.There are a few types of reccomendar systems: collaborative filtering, content-based filtering, and hybrid methods.
TikTok a social media platform that many describe as addicting is known for utilizing recommendation systems. TikTok’s “For You Page” (FYP): The FYP is TikTok’s main feature, where the app curates personalized videos for each user. It shows content based on: user interactions (likes, shares, comments), viewing patterns (how long you watch certain videos), content similarity (hashtags, trends, and video characteristics),device and account activity.
TikTok does not disclose their recommendation system but it is likely the use a hybrid approach that incorporates the following:
Collaborative Filtering for identifying user interaction patterns.
Content-Based Filtering for analyzing video metadata and features.
Deep Learning Models (e.g., CNNs, RNNs, Transformers) for processing multimedia inputs like video, audio, captions, and user interaction histories.
Contextual Bandits or Reinforcement Learning for real-time adaptability and feedback learning.
Graph Neural Networks (GNNs) to capture relationships between users, hashtags, trends, and content.
Tiktok’s target audience consist of Gen Z and Millennials but in recent years has expanded to include older age groups. I was once very reluctant to use TikTok because it was more popular amongst the youth but is now one of my favorite social media platforms because it is very informative, theres relatable contect, life hacks, DIY’s and more.
TikTok’s algorithm is unique but lets talk about something I did last week on TikTok and how it’s affected my FYP and experience on TikTok.
Last week, my partner sent me a TikTok of an engagement ring. I liked the TikTok and sent him a few videos of ones I liked. Now, a week later my FYP is primarly videos of rings. Although, I love seeing rings, I do not want to watch hundred of videos of rings.
This has happened to me with many different things for example, one time I saw this tiktok about a girl who took a soulcycle class and ended up in a hospital for a week because of a very rare diseases that breaks down muscle and causes organs to fail if not treated. I interacted with this video and now my FYP is comprised of these rare diseases which I love to learn about but now I am anxious about all of the rare dieases I can possibly expose myself too by things I do on a daily basis.
Many multicultural users have raised concerns about TikTok’s lack of representation and inclusivity. In my own experience, this has been evident as well. For instance, when I searched for curly hair styles, the majority of content shown was focused on white individuals with wavy hair rather than showcasing the full range of curl patterns and diverse hair types. This lack of representation can make it harder for users with different hair textures or styles to find relatable content and feel represented on the platform.
Increase Transparency in the Recommendation Algorithm: Allow users to understand how their data and interactions influence the content they see by providing insights into the recommendation engine’s logic. A feature or dashboard that shows users why a particular video appears on their For You Page (FYP) can help improve this.
Address Algorithmic Biases: Algorithms can inadvertently reinforce harmful patterns or biases (e.g., promoting certain body types, gender stereotypes, or content types disproportionately).Tiktok can improve this by ensuring fair representation across diverse groups (gender, ethnicity, culture, content themes).
Focus on Content Quality Over Virality: Tiktok should prioritize authenticity, sometimes TikTok promotes viral trends without accounting for whether content is uplifting, ethical, or appropriate.
TikTok’s recommendation system is a great example of how personalized algorithms can shape the way we interact with social media. By looking at user activity, preferences, and engagement, TikTok creates a For You Page (FYP) that keeps users connected to content that’s entertaining, trending, and relevant to their interests.
There’s always room for impovement like being more transparent about how recommendations are chosen, giving users more control over their feeds, making recommendations more diverse, and addressing biases in the algorithm could make the platform even better. These changes would not only improve user experience but also build trust and promote fairness. Addressing these gaps would ensure that all users feel seen, valued, and included within the community.