An Intelligent Group Recommender System using Henry Gas Solubility Optimization Algorithm-based Adaptive Deep Belief Network

Group Recommender Systems (GRS) have recently gained attention for use in various applications. The GRS model is useful in any scenario that involves a decision-making process for a group of users. However, existing approaches are not able to provide high quality recommendations due to their inability to handle traditional problems such as cold start, sparsity, and dealing with large dimensional information. To address these issues, a novel deep learning-based adaptive model is proposed. Raw data is first collected from four different data sources and cleaned using stemming, stop word and punctuation removal methods. Data preprocessing is followed by the feature extraction phase. A Text Convolutional Neural Network (TCNN) is employed to extract efficient features from group reviews. The efficient features are fed into an Adaptive Deep Belief Network (ADBN) to provide the recommendation output. It is further optimized using the Henry Gas Solubility optimization (HGSO) algorithm. Through extensive experiments it is found that the performance of the proposed system outperforms existing standard algorithms, hence, solving the traditional GRS issues.