Recommendation system of sports center in Malaysia using content-based filtering / Ahmad Fitri Zulkifli and Zainab Othman

• Main Authors: Zulkifli, Ahmad Fitri, Othman, Zainab
• Format: Book Section
• Language: English
• Published: Faculty of Computer and Mathematical Sciences 2023
• Subjects: Integer programming
• Online Access: https://ir.uitm.edu.my/id/eprint/93879/1/93879.pdf; https://ir.uitm.edu.my/id/eprint/93879/

A Sports center serves as a venue where individuals and groups can book sport courts for activities. However, the traditional method of booking directly at the venue has limitations, as it lacks recommendations for available sports courts. To address this, a new method utilizing Term Frequency-Inverse Document Frequency (TFIDF) and Cosine Similarity technique is proposed to enhance the sports center’s recommendation system. The approach is based on contents-based filtering, focusing on providing personalized suggestions to customers based on their preferences. The proposed method is implemented through a web-based platform. The study includes two types of testing: functionality testing and cosine similarity testing. Functionality testing aims to validate the system’s requirements and features, all of which are successfully met. Cosine similarity testing involves evaluating user profiles, locations, and sports courts to measure the system’s recommendation accuracy. Results from functional testing indicate the correct implementation of all requirements. The cosine similarity scores, which gauge the similarity between user profiles and locations, range from 0.2781 to 0.5031 for different test cases. Moreover, the similarity score between courts and locations is consistently 1.0, demonstrating that the recommendation details align accurately with user input and preferences. In conclusion, optimizing the sports center recommendation system through content-based filtering, specifically employing TF-IDF and Cosine similarity technique, empowers customers to choose sports centers that closely match their preferences. This system leverages the user’s history of location and court booking to provide meaningful and relevant recommendations.