DemogCF model of personalized recommendations based on demographic characteristics for overcoming data sparsity and cold start problems

With the rapid expansion of the internet, the size of e-commerce websites expanded with a huge number of products that need to be processed. This makes internet users confused with a number of choices that caused difficulty to find products suited to their preferences. Recommendation Systems (RS) are information filtering tools that help the user to find the appropriate products by generating personalized recommendations. The collaborative Filtering (CF) approach is suffering from data sparsity and cold start problems due to the absence or the insufficiency of user ratings. In order to resolve these problems, we propose in this paper a novel model, named DemogCF, that used the demographic information age, gender and occupation to generate ratings for unrated items. New ratings will be combined with the existing ratings by the Recommender System for items recommendation. In fact, filling all gaps in the ratings matrix will, necessarily, fix the data sparsity and cold start problems. Our model is compared with 2 other models on 2 standards dataset. The proposed RS provided good results in terms of standards evaluation metrics.