Publication Date

Spring 2023

Degree Type

Master's Project

Degree Name

Master of Science (MS)


Computer Science

First Advisor

William Andreopoulos

Second Advisor

Faranak Abri

Third Advisor

Manoj Desaraju


hashtag recommendations, graph-based deep learning


Many businesses, including Facebook, Netflix, and YouTube, rely heavily on a recommendation system. Recommendation systems are algorithms that attempt to provide consumers with relevant suggestions for items such as movies, videos, or reels (microvideos) to watch, hashtags for their posts, songs to listen to, and products to purchase. In many businesses, recommender systems are essential because they can generate enormous amounts of revenue and make the platform stand out when compared to others. Reels are a feature of the social media platforms that enable users to create and share videos of up to sixty seconds in length. Individuals, businesses, and organizations frequently use reels to display their creativity, advertise their products, and communicate with their audience. Users annotate these reels with hashtags that, in their opinion, properly depict the content of the reel. Therefore, it is important to consider user preferences of the content of the reel and their preferences of the hashtags to make relevant recommendations. In this project, we focus on providing users with hashtag recommendations for the reels they want to post, based on individual user's preferences of the content of the reel and hashtags, thus creating a personalized recommender for each user. Most methods that implement hashtag recommendations model interactions between users and hashtags or hashtags and posts, unlike this scenario, where we design a hashtag recommendation system based on users, reels, and hashtags. The dataset was built using web scraping, which downloaded the reels from TikTok. We designed a personalized hashtag recommender to recommend hashtags to users based on their previous posts, taking into account both their preferences for the content of the post and their understanding of hashtags. Our proposed graph deep learning based model outperformed existing approaches by achieving a NDCG score of 0.9156 which is significantly higher than the existing approaches.