Web personalized recommendation based on ontology evolution and machine learning

Resumo

The explosive growth of information available online frequently overwhelms users. Recommender system (RS) provides a useful personalized way of discovering products or services they might be interested in from a large space of possible options. Recommender system has been playing a more vital and essential role in various information access systems to boost business and facilitate decision-making process. In general, the recommendation lists are generated based on user preferences, item features, user-item past interactions and some other additional information such as temporal and spatial data. Recommendation models are mainly categorized into collaborative filtering, content-based recommender system and hybrid recommender system based on the types of input data. However, these models have their own limitations in dealing with data sparsity and cold-start problems, as well as balancing the recommendation qualities in terms of different evaluation metrics. The past few decades have witnessed the tremendous successes of the deep learning (DL) in many application domains such as computer vision and speech recognition. Recently, deep learning has been revolutionizing the recommendation architectures dramatically and brings more opportunities in reinventing the user experiences for better customer satisfaction. Recent advances in deep learning based recommender systems have gained significant attention by overcoming obstacles of conventional models and achieving high recommendation quality. Deep learning is a sub research field of machine learning. It learns multiples level of representations and abstraction from data, which can solve both supervised and unsupervised learningtasks. Thus, the recommender system plays an important role in online retail.It increase the number of items sold, Sell more diverse items which is one of the major functions of the RS that will make the users choose items that they can't easily find without recommendations, it also increase the user satisfaction, increase user fidelity which is the loyalty of the user to the website that will be achieved after the recognition of the RSs for him as an old customer and treats him as a valuable visitor, by giving him recommendations according to his past behavior over the website, e.g., his ratings for his previous items. Large information stores as digital libraries, online shops, etc. rely on a study of the objects under consideration to structure the vast contents and help in browsing and searching. As in heterogeneous communities, users usually use different parts of such ontology with varying intensity, customization and personalization. Such personal ontologies reflect the interests of users at certain times. Interests might change as well as the available data; therefore the personalization requires quite naturally support for the evolution of personal ontologies. As ontology evolution is defined by the adaption of ontology to changes that are appeared in the domain of interest . Since ontology reasoning helps in detecting the changes in user interests; ontology reasoning, then, refers to the inference of new knowledge from the statements asserted by an ontology engineer. The reasoner recognizes the illegal mistakes that are committed by the developer. Use of ontology is increasing in Information Systems and Knowledge Sharing Systems that increase the significance of ontology maintenance. Ontologies are formal description ofshared conceptualization of a domain of discourse. In the proposed thesis a social semantic personalized recommendation model will be proposed by the use of deep learning /reinforcement learning based on ontology evolution technique to obtain an updated ontology from different data sources such as, text documents, databases or other ontologies that can be tested by current ontology reasoners to make updated recommendations that are suitable to the new user’s preferences ad change of his behavior over social media.