DESI: Diversification of E-Commerce Recommendations Using Semantic Intelligence

Due to the dynamism in the Web 3.0 there is need for semantically driven framework for ecommerce based recommendations. This paper proposes the DESI framework, which is a query-driven, semantically oriented, Web 3.0 conforming ecommerce recommendation framework. The pre-process query was enriched using the Latent Semantic Indexing. Ontologies are generated from the ecommerce product dataset. The classification of the metadata takes place using the LSTM classifier. The relevance computation is staged and is achieved using Lin similarity, Adaptive Pointwise Mutual Information measure. The semantics oriented reasoning on the basis of semantic similarity measures yield the matching products and Frog Leap algorithm. The normalized point wise mutual information measure is used to compute the intermediate results and probability algorithm ensures optimality computation from the initial feasible solution set. An overall precision accuracy measure of so and so on with the lowest value of False Discovery Rate has been achieved with a proposed framework.