Aspect Category Detection of Mobile Edge Customer Reviews: A Distributed and Trustworthy Restaurant Recommendation System
Sidra Abbas, Wadii Boulila, Maha Driss, Nancy Victor, Gabriel Avelino Sampedro, Mideth Abisado, Thippa Reddy Gadekallu
Abstract
Mobile Edge Computing (MEC) enhances social media customer reviews by providing real-time processing and analysis at the network edge. Social media platforms have revolutionized how users communicate their thoughts, ideas, and opinions on various issues, yielding a wealth of valuable data that can be used to get insights into people’s attitudes toward various items. This data is especially relevant for Aspect Category Detection (ACD), which is finding specific aspects or features that people discuss or mention concerning products. These systems are used in various organizations that operate on various platforms. However, current approaches frequently need to generate promising and accurate outcomes. Thus, this research presents an innovative strategy for detecting ACD in user reviews of restaurants using federated learning. The approach harnesses the power of a federated deep neural network for accurate classification. Various data preparation methods were employed to preprocess the data before model training to ensure the construction of a reliable dataset for classification. The proposed approach includes data cleaning, balancing, TF-IDF feature extraction, and model prediction using federated learning. The experimental results demonstrate that the proposed approach achieved an impressive accuracy of 88.38%, recommending the proposed approach for aspect category detection.