MRDFPD: Metadata Driven RDF Based Product Discovery Framework

Abstract

AbstractIn recent years, the number of people who use e-commerce has risen rapidly. Users can utilise a feature on shopping sites to search for items based on their descriptions. The inputs can be thought of as product discovery and recommendation queries. Product recommendation was previously made using syntactic methods, which were ineffective in categorising and discovering appropriate products solely based on keyword matching. Each online platform has a large number of products that must be organised into taxonomies in order to improve product discovery accuracy. A metadata-driven RDF-driven framework for product discovery and classification is proposed in this paper. Preprocessing and query upper ontology matching are included in the queries. The knowledge base that stores the common categories from e-commerce sites is held by the category upper ontology, a semantic model. The proposed framework takes into account the input queries as well as the number of user clicks. It uses logistic regression for product classification on the dataset and the Harris’ Hawks optimisation algorithm to compute semantic similarities to produce accurate recommendations and rankings. KeywordsHarris’ Hawks optimizationLogistic regressionRDFSemantic similaritiesUpper ontology