Recommender systems have a better performance whenever they have more information about the users. When little to no information is available for a certain user, the system fails to provide them with accurate recommendations, and this is referred to as the cold-start problem. To solve this issue, non- personalized and personalized active learning strategies were proposed, with an aim to obtain some ratings from the new user on specific items. In this paper, we tackle the non-personalized category, we compare the strategies that were previously suggested in the literature, and we propose a new strategy that combines some selected previous strategies. We use the public MovieLens dataset in the experimentations and we use MAE and RMSE to compare the strategies. The results show that our strategy performs better than the individual strategies previously used.