Research on Grid Data Analysis and Intelligent Recommendation System by Introducing Neural Tensor Network Model

Abstract

In the landscape of modern smart homes, the prevalence of intelligent devices, notably smart televisions (TVs), has surged, emphasizing the need for sophisticated content recommendation systems. However, the automatic provision of personalized content recommendations for smart TV users remains an underexplored domain. Existing literature has delved into recommendation systems across diverse applications, yet a distinctive void exists in addressing the intricate challenges specific to smart TV users, particularly the incorporation of the smart TV camera module for user image capture and validation. This research introduces a pioneering Intelligent Recommendation System for smart TV users, incorporating a novel Convolutional Neural Tensor Network (CNTN) model. The implementation of this innovative approach involves training the CNN algorithm on two distinct datasets “CelebFaces Attribute Dataset” and “Labeled Faces in the Wild-People” for proficient feature extraction and precise human face detection. The trained CNTN model processes user images captured through the smart TV camera module, matching them against a ’synthetic dataset.’ Exploiting this matching process, a hybrid filtering technique is proposed and applied, seamlessly facilitating the personalized recommendation of programs. The proposed CNTN algorithm demonstrates an impressive training performance, achieving approximately 97.18%. Moreover, the hybrid filtering technique produces commendable results, attaining an approximate recommendation accuracy of 89% for single-user scenarios and 86% for multi-user scenarios. These findings underscore the superior efficacy of the hybrid filtering approach compared to conventional content-based and collaborative filtering techniques. The integration of the CNTN architecture and the hybrid filtering methodology collectively contributes to the development of an advanced and effective recommendation system tailored to the nuanced preferences of smart TV users in the context of grid data analysis.