A NOVEL PRODUCT RECOMMENDER SYSTEM BASED ON TREND AND PRODUCT AVAILABILITY USING HR-FR AND 3SDL-GRU

Artificial Intelligence (AI) is utilized by the Product Recommender System (PRS) to learn the heterogeneous data for proper recommendation.None of the prevailing works focused on PRS centered on User Preference Data (UPD) and trends.Hence, a Swish Scaled Stochastic Depth Lasso-based Gated Recurrent Unit (3SDL-GRU)-based PRS is proposed in this paper using UPD, browser history, and trend.Primarily, the UPD is gathered and pre-processed.After that, the dimensionality reduction is carried out utilizing Principal PolySem-kernel Component Analysis (P2SCA).Next, by employing the K-Means Clustering (KMC), the data grouping is performed.Afterward, the Afinn sentiment analysis is carried out; in addition, the features are extracted from the user's browsing history.Then, by utilizing the Harmonic Ramp-Fuzzy Rule (HR-FR), adaptive filtering is performed.In the meantime, the keywords are extracted.Then, by utilizing the Bidirectional Suffix Array-based Encoder Representation form Transformers (BSAERT), the word is embedded.Moreover, the features are extracted from the grouped data.Lastly, for product recommendation, the extracted features and the output from word embedding, sentiment score, and the adaptive filter are given to the 3SDL-GRU.Therefore, the products are effectively recommended with a 0.438 Mean Squared Error (MSE), which outperforms prevailing methodologies.