Heterogeneity Involved Network-based Algorithm Leads to Accurate and Personalized Recommendations

Heterogeneity of both the source and target objects is taken into account in a network-based algorithm for the directional resource transformation between objects. Based on a biased heat conduction recommendation method (BHC) which considers the heterogeneity of the target object, we propose a heterogeneous heat conduction algorithm (HHC), by further taking the source object degree as the weight of diffusion. Tested on three real datasets, the Netflix, RYM and MovieLens, the HHC algorithm is found to present a better recommendation in both the accuracy and personalization than two excellent algorithms, i.e., the original BHC and a hybrid algorithm of heat conduction and mass diffusion (HHM), while not requiring any other accessorial information or parameter. Moreover, the HHC even elevates the recommendation accuracy on cold objects, referring to the so-called cold start problem, for effectively relieving the recommendation bias on objects with different level of popularity. Introduction The development of internet has made it easy to access information, which also brought about great convenience for our daily life. On the other hand, when facing various information, one is also puzzled how to get what he/she really wants. As a powerful tool, recommender system emerges to help people out of the overloaded information, which therefore attracts great interest of scientists from different disciplines [1], including physicists [2, 3]. Different algorithms have been proposed, and achieved considerable progress. One of the most widely applied algorithm is the so-called collaborative filtering algorithm [4, 5], which can also be divided into the memory-based [6–8] and model-based collaborative fitering [9–11]. Another Line is the content-based algorithm [12]. Various extensive algorithms have been comprehensively investigated [13–17], such as the hybrid algorithms of collaborative filtering and content-based algorithm. A recommender system can be taken as a complex system composed of some interactive units [18–26]. The complex interactions may affect the user’s activity. For example, it makes a movie well known by advertising for it via medias like internet or television. After watching the movie, audiences may feedback favorable comments or negative criticism, which on the other hand should have an impact on the potential audiences, as well as the popularity of the movie. The comments made by users to an extent reflect the individual preference. Now, it has accumulated a large amount of historical data of the users’ past activities, which makes it possible to design effective recommendation algorithms, and provide personalized recommendation for users by analyzing the data.