Enhancing Service Recommendations in the Social Internet of Things: An Adaptive Collaborative Filtering Approach Using Friendship-Based Similarity
نویسندگان
1 PhD Student. Department of Computer Engineering, Yazd University, Yazd, Iran.
2 Assistant prof. Department of Computer Engineering, Yazd University, Yazd, Iran
doi
10.22091/jemsc.2025.11431.1210چکیده
The Social Internet of Things (SIoT) integrates social interactions with IoT technology to create intelligent, connected environments. In this network, people form friendships, and objects owned by them provide services to others. As the SIoT network grows, offering personalized services tailored to individual interests becomes increasingly important. This paper examines real-world data from the city of Santander, analyzing the number of users, friendship degree distribution, and its patterns. An adaptive consonance filter algorithm based on friendship communities (A-CFA-FC) is proposed, which uses friendship relations and individual preferences to identify friendship communities based on a similarity index. The algorithm ranks and recommends services according to user interests within the SIoT environment. Results from the Santander dataset show that the proposed algorithm, using the Jaccard similarity index, detects more communities with lower time complexity and higher compactness. Compared to the baseline algorithm, it reduces root mean square error by about 17% and improves the F1 score by approximately 21%.
