An improved HeatS+ProbS hybrid recommendation algorithm based on heterogeneous initial resource configurations

摘要

Network-based recommendation algorithms for user-object link predictions haveachieved significant developments in recent years. For bipartite graphs, thereallocation of resource in such algorithms is analogous to heat spreading(HeatS) or probability spreading (ProbS) processes. The best algorithm to dateis a hybrid of the HeatS and ProbS techniques with homogenous initial resourceconfigurations, which fulfills simultaneously high accuracy and largediversity. We investigate the effect of heterogeneity in initial configurationson the HeatS+ProbS hybrid algorithm and find that both recommendation accuracyand diversity can be further improved in this new setting. Numericalexperiments show that the improvement is robust.