Tree-Based Contextual Learning for Online Job or Candidate Recommendation With Big Data Support in Professional Social Networks

Enabling users conveniently accessible to their most interesting jobs or candidates is an important and challenging task for professional network recommenders in professional social networks (PSNs). Nowadays, PSNs accommodate tremendous and diverse users and face continuous uploading and high data freshness. In this paper, an online mining and predicting system is proposed for personalized job or candidate recommendation with big-data support. It considers the users' explicit information as context to achieve personalized recommendation. In addition, the system utilizes the reward extracted from online implicit information of the previous users with similar context information to relieve the cold start problem. Furthermore, a tree-based method is introduced to address large-volume items by effectively analyzing them in cluster level and thus reduce the computational load. Considering the dynamic property of PSNs, our model is adaptive for the expanding dataset, enabling it to real-timely make accurate recommendations for the incessant new arrivals. Moreover, theoretical analysis shows that our algorithm achieves sublinear regret over time. Finally, extensive experiments are conducted to test our algorithm based on the real dataset from ACM RecSys Challenge
¹
and validate the outstanding performance of our algorithm when compared with other existing algorithms.