Bushra Alhijawi | Princess Sumaya University for Technology (PSUT) (original) (raw)

Uploads

Papers by Bushra Alhijawi

Recommender system is a helpful tool for helping the user in cutting the time needs to find perso... more Recommender system is a helpful tool for helping the user in cutting the time needs to find personalized products, documents, friends, places and services. In addition, the recommender system handles the century web problem: information overload. In the same time, many environments or technologies (i.e. cloud, mobile, social network) become popular today and facing the problem of large amount of information. Therefore, the researchers recognize that the recommender system is a suitable solution to this problem in those environments. This paper, reviews the recent research papers that were applied the recommender system in mobile, social network, or cloud environment. We classify these recommender systems into four groups (i.e. mobile recommender system, social recommender system, cloud recommender system and traditional (PC) recommender system) depending on technology or environment that the RS is applied in. This survey presents some compression, advantages and challenges of these types of recommender systems. Also, it will directly support researchers and professionals in their understanding of those types of recommender systems.

The social-based e-commerce as we named is a type of e-commerce, appears as a result of widesprea... more The social-based e-commerce as we named is a type of e-commerce, appears as a result of widespread of social networks like Facebook, Instagram, YouTube, and Twitter. The retailer in social-based e-commerce depends on the personal profile, pages, or groups on the social networks to presents their products. In this study, we investigate the factors that affect the customer satisfaction and e-loyalty in this type of e-commerce. Four factors are proposed (i.e. usefulness, usability, customer services, and customization) that may have impact on customer satisfaction and may derive the e-loyalty in social-based e-commerce. Online survey was distributed and 95 complete and valid questionnaires were returned. All the four factors have a positive impact on customer satisfaction. In addition, the customer satisfaction has a positive impact on e-loyalty in social-based e-commerce. The findings suggest that to determine e-satisfaction in SEC, the merchants need to focus on providing a handmade , tailor-made, variant colors and sizes products, special attention by the salespersons and deliverymen to the customers. This will reflect on the loyalty of the customers.

—Recommender systems aim to help web users to find only close information to their preferences ra... more —Recommender systems aim to help web users to find only close information to their preferences rather than searching through undifferentiated mass of information. Currently, col-laborative filtering is probably the most known and commonly used recommendation approach in recommender systems. In this paper, we present a new genetic algorithms-based recommender system, SimGen, that computes the similarity values between users without using any of the well-known similarity metric calculation algorithms like Pearson correlation and vector cosine-based similarity. The results obtained present 46% and 38% improvements in prediction quality and performance, respectively when compared with other techniques.

Thesis Chapters by Bushra Alhijawi

The virtual world overflowing with the digital items which make the searching, choosing and shopp... more The virtual world overflowing with the digital items which make the searching, choosing and shopping hard tasks for users. The recommender system is a smart filtering tool for generate a list of potential favorite items for the user to reduce the time needed by user to
choose among a huge number of choices in websites and facilitate the process.
In that context, this thesis presents a novel technique that combines the ideas of item-based semantic similarity, n-criteria and multi-filtering criteria with the genetic-based recommender system. The genetic algorithm is utilized in order to predict the best list of items to the active user. Consequently, each individual in the population represents a candidate recommendation list. Each list subjects to three tests to measure the quality of it.
The proposed system alleviates the effect of the sparsity and cold start problems and makes the recommender system capable of generating recommendation without the need of using a similarity metric or requires any additional information provided by the hybrid system. Furthermore and due to the fact that there are many environments facing the information overload problem, the author presents a new classification of the recommender system based on the environment that is applied in.
The proposed system is evaluated against the state-of-the-art genetic-based recommender system and the traditional techniques that used in collaborative filtering recommender system. The results obtained show that the proposed method outperforms these algorithms in prediction accuracy by 24.3%, recommendation quality by 33.5% and performance (CPU time) by 45.4%. Moreover, the results showed that 69.5% of the recommended items are truly favorite items to the active user. The remainders 30.5% of the recommended items are potential favorite items.