A Survey on Finding Novelty in the Recommender Systems by Adding Lower Similar Items in the Top-N List

The Internet has made it very challenging to find useful information from the vast amount of online data. Recommender systems aim to suggest items that are both useful and unexpected to users. These items are beneficial for the retailers and also surprisingly fit the consumers' preferences. However, existing methods struggle to provide novelty due to the skewed distribution of observed data for popular and tail items. User satisfaction with recommender systems depends on how well the system matches the user's needs and how much it helps the user make decisions. Accuracy alone is not sufficient, because the quality of recommendation also matters. The quality of recommendation is defined by how well it meets or exceeds the customer's expectations, and the user will not be happy with some of the repeated and “not so interesting� suggestions. We reviewed the papers to understand how to introduce novelty in recommendations to improve user satisfaction.